Home page logo

nanog logo nanog mailing list archives

Re: East Coast outage?
From: David Lesher <wb8foz () nrk com>
Date: Sat, 16 Aug 2003 14:39:37 -0400 (EDT)

Speaking on Deep Background, the Press Secretary whispered:

AC still makes sense for distribution, but HV DC transmission lines are
becoming the norm.  Think about some very large SCRs and associated parts
to convert to AC for distribution.

For several reasons....

You must size the equipment inc. insulators for the peak voltage
(plus whatever safety factor, of course...). With DC, the peak
voltage is by definition equal to the RMS [Root-Mean-Square]
voltage [0]. For sinosoidal AC there's this SQR 2 factor you must
throw in. Thus, given a line insulated with say 18" insulators,
you can run a higher DC voltage and thus carry more power.

Secondly, there are AC capacitive losses. Power lines, especially
buried ones, have some capacitance to ground & each other. In
a AC system, that's a loss; in a DC it's actually storage... If
you walk under a 132/354KV line on a humid day, you can hear the
hum. If you hear it, it's wasted energy.

But a very important reason for choosing DC is a good NANOG
term: Asynchronous. 

With a long AC line, both ends must be exactly in sync. Phase
differences actually control which end is the source, and which
the load. [I can remember Prof. Klingshirn proving this on the
blackboard but no way would I go any deeper than this..]  So you
end up with some REALLY nasty non-linear effects that resemble
trying to use an screen editor over a slow link with both variable
latency and severe packet loss. The difference is, here things
blow up like an Irwin Allen set; instead of just turning "Hello
World" into "^H%$^&%EBBFFSJJS"

When things get more out of phase, cutouts trip to CYA. If they
don't trip fast enough, really big expensive rotating machines
compensate for that by breaking loose from the floor and playing
pavement roller. And if they trip, the lights go out. Do you
wait 150 ms and see if things quiet down, or save your hide by
jumping ship?  It's more like paying poker than blackjack.

DC solves the whole phase issue: the energy direction is a voltage
difference function, not phase. That's why it's used to bridge
different power grids.

But the price is larger initial costs. DC makes circuit breakers
(interrupters) far harder to build than AC. You need massive SCR
banks and diode arrays; they have their own losses inherent in
their operation. Further, we have many decades of engineering
experience with HV AC and far far less with +100KV DC.

Lastly there is separate major major non-engineering issue. Up
until recently, the national grid sorta resembled the pre-84
Bell System. Sure there were separate companies but they were not
in direct competition with each other. Now it's almost like our
telecom/peering backbone system, and you have the same economic
Prisoners Dilemma. While it might, in a dispassionate stability
analysis, make the most sense to dump all of the NASA Pathfinder
traffic onto {say} AGIS or PSI and let them eat it; they will
likely disagree and do the opposite. Retail suppliers like ConEd
or PEPCO clearly want to buy the cheapest power whenever they can
[1], and sell it to their users. But to get it to DC from Canada,
it must transit lines now owned by a third party, who wants to move
the most they can (and thus get paid the most, of course..) and
thus runs them as hard as they can. Harder == less slack to handle
a sudden change, be it from your load, or a Disturbance In The Force 
somewhere 750 miles away.

Just as security is the opposite of efficiency; stability and
redundancy are the enemies of maximum profit. So where's the
bright line set? Good Question [TM]. 

Maybe those Y2K/Montana Milita types are right; get off the
grid anyhow because the flouridation of the network is a massive
Commie^^H^H^H^H Saudi^H^H^H Freedonian plot. I donno. I do know
it is not going to get better, and any White House program to
"fix the problem" will make it 3x as bad.

[1] and they likely buy it from a broker/reseller like Enron, further
muddying the waters..

[0] VRMS is the AC voltage that if you run it though a resistor,
you get the DC-equal watts of heat. Take 10v DC and and a one ohm
resistor, and you get 100 watts. Same for 10 V RMS as well. Not
the same for 10 VAC "average" "peak" "peak-to peak" or whatever --
they'll all differ with waveform. In fact only really accurate
VRMS meters used to be lab-precision resistors and calorimeters.

A host is a host from coast to coast.................wb8foz () nrk com
& no one will talk to a host that's close........[v].(301) 56-LINUX
Unless the host (that isn't close).........................pob 1433
is busy, hung or dead....................................20915-1433

  By Date           By Thread  

Current thread:
[ Nmap | Sec Tools | Mailing Lists | Site News | About/Contact | Advertising | Privacy ]