[time-nuts] 60 Hz measurement party
phk at phk.freebsd.dk
Sun Jun 26 17:38:54 UTC 2011
In message <4E066FAC.5090709 at rubidium.dyndns.org>, Magnus Danielson writes:
>Infact, Poul-Henning and I had the idea to test this on our grid to see
>what kind of performance we would get out of it. He sent me a
>transformer prepped for the work, but it seems both of us got caught up
>doing other stuff to follow through, but this is a good trigger.
Yeah, life, don't talk to me about life... :-)
Actually what I wanted to measure back then was the phase-stiffness
of the grid between us.
This may be relevant to Tom's experiment as well, and in particular
to those of you living in California, so let me explain what it is:
Imagine two time-nuts, N1 and N2, two generators G1 and G2 and
various chains of transformators T1a... and T2a... in the same
powergrid but some n*100km apart.
G1 --- T1a -+- T1b ------------------------T2b -+- T2a --- G2
If all the power is generated by G1, the absolute GPS relative
phase seen at N1, depends on G1, T1a and T1c, which N2 sees
the combined effects og G1, T1a, T1b, T2b, T2c. When G2 produces
all the power, the picture reverses and when G1 and G2 each produce
half the phase difference btween N1 and N2 should be constant.
(But not zero, because they may not be on the same of the three
phases and because grid transformers shifts phases around to match
things up. Long story, not for today.)
Which way you push power through T1b and T2b affects what they do
to the phase of the grid on either side so the phase difference
seen between N1 and N2 depends on how power flows in the grid.
The reason everybody in the same grid sees the same frequency, is
that when you have a big heavy generator, frequency is a usable
proxy measurements for energy.
If you add an electrical load, the generator have to produce more
electricity which takes more mechanical work causing the
turbine to slow down. And vice versa.
Typically, a frequency deviation of as little as 0.02 % will cause
regulation of turbine steam.
Load changes also cause the voltage to change, but this is much
less pronounced and much harder to measure/regulate with, primarily
because of the very noisy measurements.
So the power-grid basically doesn't use voltage for regulation.
Various mechanisms keep the voltage inside a +/- 10% tolerance
at various points and that's that.
With me so far ?
All this breaks down once we start adding power-producers which
are not based on huge rotating lumps of copper-ensnarled iron.
Solar cells, wind generators, HVDC transmission, electrical cars
feeding battery power and all these other fancy modern things, feed
power into the grid with a computer controlled switch mode gadget
which just tracks whatever phase and frequency your grid has right
When the frequency changes on one of these switchmodes, they just
follow the grid, they do not try to join in on the "voting" on
the frequency by trying to pull the grid ahead or behind depending
on their power-state.
As the grid moves from big rotating lumps of iron to switch mode
attachment, a larger and larger fraction of the generation capacity
free-wheels in the frequency 'voting'.
At some point, the system will no longer be stable, and something
has to happen.
My particular corner of the world is ground-zero for this, because
we generate 1/5th of our electricity with windmills and have
relatively little rotating machinery running in good winds.
So far, we are not approaching instability, at least not so that
anybody will admit it.
But the way to tell if instability is approaching, is to monitor
the phase difference between N1 and N2 as explained above, the
larger variations and the more resonance frequencies manifest
themselves therein, the more you should get involved in local
The future of grid regulation is to move to a "absolute frequency"
model, where the frequency is UTC-locked through-out the grid, and
regulation happens only on voltage.
There is a lot of fancy technology involved in this, and som scary
propositions about what we can and can not do with "holistic grid
regulation" and other such buzzwords.
Really long term, Edison will win and long-haul electricity will
all happen on HVDC lines. When we get buck/boost converters working
directly on HVDC, everything will be much simpler and stabler, so
people are seriously dragging their feet.
My idea for measuring this, was to measure the time from the
utc second from a GPS receiver to the first zero-crossing of
the grid, and try plot Magnus and my measurements together.
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
More information about the time-nuts