San Francisco's secret electric power grid

November 29, 2013 11:36 AM

"While DC continues to race through San Francisco power lines at nearly the speed of light, it does so anonymously. You’ll find no reference to DC power distribution in PG&E’s annual reports or on its websites. Even some utility engineers are unaware of its existence, which raises a curious question: Why is the inheritor of this legacy, the mighty and sophisticated PG&E, still bothering with DC distribution 133 years later?"

With thanks to mollymayhem for pointing me to the article.

posted by gingerbeer (44 comments total)

Give it up, still living head of Thomas Edison hooked up to machinery somwhere under Menlo Park.
posted by Artw at 11:43 AM on November 29, 2013 [13 favorites]

This part: However, safety was a weak point. If a winding drum’s control system fails, its motor can drive the elevator through the roof, according to San Francisco–based elevator consultant Richard Blaska.

makes me want to know which buildings still have these elevators, so that I can never, ever ride them.
posted by rtha at 11:57 AM on November 29, 2013 [3 favorites]

Interesting article, and I had no idea DC distribution was still a thing.

One pedantic note, however: Electric current, as defined by electron flow, is slow - on the order of a few meters per hour. What happens at nearly the speed of light is the propagation of changes to that current. It's like turning on and off the water supply to a long garden hose at the source and seeing the resulting starts and stops of water almost instantaneously, even though the actual speed of the water molecules in the hose is much slower.
posted by rocket88 at 12:08 PM on November 29, 2013 [19 favorites]

Here in Europe there are big plans to introduce High voltage DC (HVDC) cables (a European Super Grid) allowing electricity to be transmitted over much greater distances than existing AC lines. The £500m BritNed cable, which has just entered operation, runs from the Isle of Grain in Kent UK to Maasvlakte, near Rotterdam, in the Netherlands.
posted by Lanark at 12:14 PM on November 29, 2013 [5 favorites]

Neat, had no idea that DC grids still existed.
posted by octothorpe at 12:16 PM on November 29, 2013

I remember back in the late 1980's, there was still DC in parts of Boston - particularly used for elevators in some buildings, iirc. They were still taking it out then, but I assume they've finished getting rid of it by it now.

As for those elevator motors, I think the workaround was to keep the elevator gear as-is and just install on-site rectifiers.
posted by rmd1023 at 12:20 PM on November 29, 2013 [1 favorite]

This is shocking news!
posted by notme at 12:20 PM on November 29, 2013 [1 favorite]

That was really interesting. And the conclusion is food for thought:

Could holdouts such as San Francisco ultimately have the last laugh? The spread of both DC generators (such as photovoltaic panels) and DC loads (such as cellphones, flat-panel televisions, LED lights, and even electric cars) is inspiring a small but growing niche for DC microgrids that link the two together. If such building-wide circuits grow into neighborhood grids and, ultimately, meld together to form citywide DC grids, this circuit of electrical history will finally be closed.

posted by Kevin Street at 12:21 PM on November 29, 2013 [2 favorites]

One of these days, I'm going to take a magnetometer under one of the Manitoba Bipole lines. 1800 amps should be a pretty substantial magnetic source.
posted by irrelephant at 12:28 PM on November 29, 2013 [1 favorite]

Also to be found, at much lower voltages, in server farms, where small increases in efficiency count for quite a lot. Cars and trucks have always been DC, although there's a move to 48v from 12v in cars while trucks are 24v, while larger aircraft and boats are AC at 400 Hz (saves weight due smaller transformers, and is why there's that characteristic faint whine on older PA and radio messages). Many exceptions to all of the above...
posted by Devonian at 12:30 PM on November 29, 2013

Also to be found, at much lower voltages, in server farms

And telcos. Running gear off -48V is much more efficient than converting AC to DC in each individual piece of equipment, and also makes backup battery power much simpler because you eliminate a DC-to-AC conversion stage at the UPS itself.
posted by mrbill at 12:36 PM on November 29, 2013 [2 favorites]

In addition, who doesn't love the old Mercury-arc rectifiers?
posted by mrbill at 1:04 PM on November 29, 2013 [7 favorites]

Isn't it the case that AC's historical advantage over DC for power distribution was solely the ease of making AC transformers, and that since the development of efficient DC-DC transformers, AC is actually less efficient, and persists mostly by virtue of the standards being a century or so old?
posted by acb at 1:19 PM on November 29, 2013

I know nothing of electrical engineering, but from the article above it sounds like a DC grid would have to be designed fundamentally differently from the AC ones. The old DC grid in San Francisco was dangerous and clunky because it tended to flow around damage (much like the Internet), which made it hard to know when something was broken or where the damage was located. I guess doing DC on a large scale would involve a whole lot of reinventing and rethinking established paradigms.
posted by Kevin Street at 1:24 PM on November 29, 2013

It puts me in mind of this recent piece - interesting how new applications of pretty much existing technology make it easier to do things that, fairly recently, were considered too difficult and/or expensive to do at any kind of significant scale.
posted by nickmark at 1:27 PM on November 29, 2013 [4 favorites]

Quebec is big into HVDC, with 735 kV systems thought up by the eggheads at Varennes. They also have DC links to neighbouring ISO/RTOs so they can maintain their own grid sync. This is why they stayed up during the blackout in 2003.

DC arcs are a bit scary, though. You can make cruddy but functional tack welds with the leads of a solar panel. Per-panel solar microinverters eliminate the DC wiring completely.
posted by scruss at 1:41 PM on November 29, 2013 [1 favorite]

I love these stories about weird old infrastructure that is maintained and functional well after it has been forgotten in the rest of the world.

Other examples from San Francisco are (of course) the cable cars, but also the Fire Department's practice of manufacturing its own wooden(!) ladders, and the antique emergency call box network which uses telegraph(!) technology, has remained in operation for well over a century, and is credited with working during the 1989 earthquake when other lines of communication failed.
posted by alexei at 1:45 PM on November 29, 2013 [13 favorites]

That's really neat, nickmark. Using USB PD cables you could set up a little power grid in your house or apartment building, and plug sources like solar panels in with little trouble. (Compared to hooking solar up to the citywide AC grid, that is.) It would look messy, with cables running everywhere instead of being tucked inside the walls - but it would be more efficient, and maybe even intelligent, if devices receive electricity only when they need it.
posted by Kevin Street at 1:45 PM on November 29, 2013

Yeah the thing about transformers is that they not only allow for convenient voltage conversion, they isolate the parts of the grid from one another so that damage is localized.

And there is no such thing as a DC-DC transformer. There are DC-DC converters which almost always work by converting the DC to AC and using an AC transformer, or by using an inductor buck-boost scheme like switching power supplies, by interrupting the power flow to alternately charge capacitors in one configuration and discharge them in a different one for higher or lower combined output voltage. The buck-boost and capacitor switching versions don't provide the kind of isolation actual AC transformers do.
posted by localroger at 1:47 PM on November 29, 2013 [1 favorite]

The 735 kV stuff is AC. HQ's DC lines are ± 450 kV. Some of that expertise came from a "right-to-landscape" lawsuit that meant Hydro couldn't build an overhead line over the St. Lawrence. They had to bury it in a tunnel under the river, and the only viable option was HVDC. Hydro did, however, build the HVDC link between Connecticut and Long Island.
posted by Monday, stony Monday at 1:48 PM on November 29, 2013 [1 favorite]

NY's Con Ed discontinued DC service in 2007:

Con Edison’s Electric Operations Manager Fred Simms, an active employee for 52 years, cut a ceremonial cable on 40th Street just east of Fifth Avenue today to retire the utility’s direct current (DC) service. The supply of DC service to New York customers dates back 125 years, to the advent of Thomas Edison’s first electric generating station on Pearl Street in 1882.

...

In January 1998, Consolidated Edison began a program to eliminate DC service in its operating territory. At that time there were over 4,600 DC customers. By 2006, sixty remained. Between then and today, when the last customer at 10 East 40th Street was switched to rectifiers on their side of the meter to generate direct current to supply its building’s elevators and sprinkler system, Con Edison has been switching DC customers to alternating current.

press release
posted by Brian Puccio at 1:51 PM on November 29, 2013 [2 favorites]

And the big thing with HVDC is that you can run in underwater. There's a big link between the North Island and the South Island of New Zealand, linking the Hydro Stations on the less populated South Island to the higher-consuming North Island. There are also a bunch of lines in Europe, in the North Sea and the Baltic Sea.

It wouldn't be too safe to use a mess of USB cable to power any kind of large load; that would be a fire hazard. Instead, you could install new cables in the walls for DC, with appropriate protection through breakers or fuses.
posted by Monday, stony Monday at 1:59 PM on November 29, 2013

This was all very close to home in my last job, where I was running a 22 story 1911 clock tower with a pair of 1911 elevators that ran DC motors because AC motors that could run an elevator didn't exist in 1911 and the city of Baltimore had a system of AC-to-DC substations precisely for industrial DC. In my building, the traction elevators ran off DC and do to this day, despite the DC grid in Baltimore being shut down in the twenties.

For each elevator, there's a huge AC motor connected to a DC generator and, when you press a button for the elevator, wwwwhhhooooooOOOOOOOOMMM, the generator fires up and powers the elevators. It's not the simplest system for running those (a proper DC grid is the best), and there's maintenance and the delay time of firing and shutting down generators all day, but the alternative is to gut the building and put in wretched modern elevators that destroy the surviving 1911 car with its openwork and gilding and brass gates.

Viva modern world and all, but I think we ought to hold on to the rare old stuff while we can.
posted by sonascope at 2:08 PM on November 29, 2013 [9 favorites]

Isn't it the case that AC's historical advantage over DC for power distribution was solely the ease of making AC transformers, and that since the development of efficient DC-DC transformers, AC is actually less efficient, and persists mostly by virtue of the standards being a century or so old? posted by acb at 3:19 PM

When I went to electronics school 25 + years ago, we were taught that the reason AC was chosen over DC was because it was easier to transmit AC over long distances with less loss. Even though transmission lines are made of the most conductive materials that are affordable, the wires do have some resistance, so there are always some losses in transmission. The formulas for power* are:

I * R = P, where
I = current
R = resistance
P = power

also

E * I = P, where
E = voltage
I = current
P = power

Using transformers, it's easy to step AC voltage up or down, but increasing DC voltage is more difficult. If power and resistance are held constant, the energy lost in transmission lines can be reduced by stepping AC voltage way up with a transformer before transmitting it, then stepping it back down when it reaches its destination. Stepping voltage up decreases current proportionally, and stepping voltage down increases current proportionally. Decreasing I in the first formula above decreases power lost. Since stepping voltage up decreases current proportionately, power remains constant in the second equation.

* The formulas given are for DC. AC formulas are a little different, but the relationships between voltage, current, resistance, and power are the same. For our purposes of understanding here, the DC formulas are close enough.
posted by Daddy-O at 2:23 PM on November 29, 2013

Electric current, as defined by electron flow, is slow - on the order of a few meters per hour.

Holes, on the other hand, move at nearly the speed of light.
posted by Chocolate Pickle at 3:31 PM on November 29, 2013

When my dad was still alive, he worked for the Army Corps of Engineers in the PNW, on the hydro projects in the Columbia basin. He told me that there was a period in the 1960's when there was an excess of power generation, so they built a transmission line from The Dalles Dam down to Los Angeles to ship excess power south, in the summer. And originally it was designed to ship HVDC.

Day came they were ready to test it. Guy in Los Angeles got on the phone to the guy in The Dalles and said "Turn it on!"

Guy in The Dalles threw a big mother switch.

Guy in Los Angeles said, "What are you waiting for? Turn it on!" Nothing was coming out at LA.

It seemed they had managed to make the transmission line a quarter wavelength (I think it was), and they'd set up a standing wave. He said the coronas around the wires at about the half-way point were legendary.

So they had to redesign to ship DC on that transmission line.
posted by Chocolate Pickle at 3:40 PM on November 29, 2013 [4 favorites]

But now that we have solid state DC voltage conversion, it's really the installed base that keeps transmission AC, isn't it? IIRC, a solid state DC-DC voltage conversion is more efficient than a standard AC transformer. That was not true back in the days of mercury valves and mechanical conversion. Or am I remembering wrong and it's that the cable loss is lower because DC doesn't have the skin effect causing much of the conductor to go unused?
posted by wierdo at 3:43 PM on November 29, 2013

> "If a winding drum’s control system fails, its motor can drive the elevator through the roof, according to San Francisco–based elevator consultant Richard Blaska. "

> makes me want to know which buildings still have these elevators, so that I can never, ever ride them.

It sounds like California's regulatory structure has made it impossible to convert the old elevators. They must be maintained as-is or replaced completely.

Article by the same Richard Blaska.

--

Another article about DC by Peter Fairley. He failed to persuade me that the fact that some generation-tier distribution is DC and charging laptops and phones (and running LED lightbulbs) can be done with DC will have an effect on the rest of the distribution system.

--
> It wouldn't be too safe to use a mess of USB cable to power any kind of large load

USB Power Delivery (PD) is up to 100 Watts (20V @ 5A) in a single cable, so you don't need a bunch of 2.5 Watt cables (4.5W for USB 3, 7.5 Watts for USB Battery Charging).

Cables capable of more than 1.5A identify themselves.
posted by morganw at 3:50 PM on November 29, 2013

>Electric current, as defined by electron flow, is slow - on the order of a few meters per hour.

>Holes, on the other hand, move at nearly the speed of light.

Nope, holes have much lower mobility than electrons. That is why P-FET devices are slower and have higher resistance than equivalent N-FETs. Holes are a charge diffused over many atoms and therefore have a larger effective mass and move more slowly than electrons under the same electric field.
posted by JackFlash at 4:11 PM on November 29, 2013 [1 favorite]

You will here more about this in the coming years as fast charging of electric cars with high voltage DC becomes a thing. As an example, BMW is advertising that their new i3 model will fully charge in about 3 hours with 240 Vac, as you may have in your home to run a clothes dryer, or about 30 minutes with high voltage DC. That's not as fast as refueling with gasoline, but it's getting there.
posted by LastOfHisKind at 4:14 PM on November 29, 2013

(As an aside, the photo at the top of the artlcle about the new cable from the Netherlands to GB makes me sad. Things have almost certainly changed since I was last in Europe or GB, but it illustrates one thing that is done so well: Villages and small towns occupy a contained, dense, footprint, and then what amounts to countryside starts right away. Here in the states, every little burg has umpteen miles of shitty tin buildings, crappy lots full of forgotten equipment, and endless strip malls and housing developments before you get to the town proper, if you can even identify it. Not only are the aesthetics depressing in the states, it's really inefficient use of land--or non-use of land, which is important, too.)
posted by maxwelton at 4:35 PM on November 29, 2013 [6 favorites]

The problem with fast-charging cars is that the most power-hungry devices in homes to date have been electric dryers, ranges, and furnaces; these might use 10 kilowatts on a circuit, and home power distribution is scaled to provide for that.

But to fully charge a Tesla 85 kwH battery at that rate takes 8 hours. To charge your Tesla in half an hour requires 170 kilowatts. At 240 volts that's 700 amps, which means copper conductors the diameter of golf balls. To get the copper conductors down to a reasonable size, carrying say 40 amps, you need about 4,000 volts. This is actually in the range of what residential distribution lines typically carry, but there is no normal way to get that voltage to individual houses; it's always stepped down to safer levels before dropping from the poles.
posted by localroger at 5:27 PM on November 29, 2013 [1 favorite]

Oops...

"And originally it was designed to ship HVDC." No, high voltage AC. Sorry for the mistake.
posted by Chocolate Pickle at 6:17 PM on November 29, 2013

I'm pretty sure that Con Ed provided DC to a few customers in NYC up until a few years ago (they also still provide heat to most of the city from a handful of centralized cogeneration plants, which is another cool oddity of NYC's utility infrastructure).

I'd expect this sort of archaic infrastructure to exist in NYC, but not in San Francisco, which is a much younger city that's burned down a handful of times. Heck, NYC powered the subway with rotary converters up until a few years ago.

The strangest electrical anachronism still in operation today is probably the 25Hz power system that powers Amtrak's Northeast Corridor.
posted by schmod at 8:36 PM on November 29, 2013 [1 favorite]

That's strange. I'm trying to think of why someone would want to use 25 Hz. My first guess is that it's easier to synchronize with a slower rate.
posted by Chocolate Pickle at 9:09 PM on November 29, 2013

Man, the comments section of IEEE Spectrum articles are like the exact opposite of my local newspaper's.
posted by Harvey Kilobit at 9:28 PM on November 29, 2013 [5 favorites]

That's strange. I'm trying to think of why someone would want to use 25 Hz.

The trains themselves used DC traction motors which were easier than AC for regulating speed and power of the train. The power distribution system used high-voltage AC for more efficient long distance transmission. At intervals along the track, the AC power was tapped and a rotary converter used to convert the AC to DC for use by the train.

A rotary converter is a combination AC motor and DC dynamo that essentially rectifies AC into DC. These are huge rotating electrical machines. The speed, number of poles and efficiency of the rotary converter are related to the AC frequency. A lower frequency reduces the speed at which the machine needs to rotate, reducing the mechanical stresses, and reducing the number of poles reduces voltage stress and increases efficiency.

So the reason for using 25 Hz AC was that the DC rectification technology available at the time, the rotary converter, favored a lower frequency than 60 Hz. This was before the development of high-power mercury valve rectifiers and of course long before power semiconductors.
posted by JackFlash at 11:16 PM on November 29, 2013 [3 favorites]

wierdo: "IIRC, a solid state DC-DC voltage conversion is more efficient than a standard AC transformer."

Your standard AC transformer is around 97% efficient; beating that is going to be hard and the margin won't be much. And you need to swap that DC voltage to AC eventually which will add further losses. Of course you only need reasonable efficiency in transformation as long as there are sufficient gains from transmission efficiency.

I'm a little surprised anyone is still using rotary conversion for relatively small loads like elevators that could be easily handled by solid state rectification. I wonder what the advantage is or if it's purely an installed capital base thing.
posted by Mitheral at 12:12 AM on November 30, 2013

Mitheral: "I'm a little surprised anyone is still using rotary conversion for relatively small loads like elevators that could be easily handled by solid state rectification. I wonder what the advantage is or if it's purely an installed capital base thing."

An office building I used to work at still had elevators controlled by a 1960s era relay controller and was fed by a pair of rotary converters until the building's new owners finally replaced them a few years ago. The age of the system and difficulty sourcing spare parts led to increasingly frequent breakdowns. The efficiency of the rotary converters wasn't really a factor, nor were they the problem.

It's not cheap to replace all the electrical bits on an elevator. There are a lot of them, and it takes a couple of months. If you've only got two (or worse, one!) elevator, that can be a serious issue.
posted by wierdo at 12:32 AM on November 30, 2013

I wonder whether a DC grid would be more vulnerable to a Carrington event or a NEMP than the AC, or less?

I'd think more, because of the way transformers filter out DC, but I'm not at all sure.
posted by jamjam at 8:45 AM on November 30, 2013

From PG&E:

7. New direct-current (d-c) or two-phase service is not available. Direct-current
service and two-phase service is supplied only to existing customers who
continue to operate existing d-c or two-phase equipment. Such service is being
gradually replaced by standard alternating-current service.

I am an engineer involved in a San Francisco project that recently pulled out an existing DC service from an existing building:

DC distribution panel

Panel + DC meter that looks more like a gas meter

Please excuse the badly lit photos, this was located in a basement and the main AC electric feed to the building had been scavenged, there was zero natural light available.
posted by vaportrail at 12:21 PM on November 30, 2013 [6 favorites]

This is so cool! Mercury valve rectifiers look exactly as you would expect of futuristic sci-fi gadgets designed in the 1950s, and their replacements oh my! Thyristor banks look like something from a Culture novel! Squeee oh how embarrasing.
posted by ianso at 12:30 PM on November 30, 2013 [2 favorites]

How cool, vaportrail. Thanks for adding those photos.
posted by gingerbeer at 3:13 PM on November 30, 2013

Why can't those buildings install a rectifier to provide DC power to the elevator?
posted by mike3k at 10:59 PM on November 30, 2013