Sharing power quality data

Last year, about this time, I asked if there was any interest in sharing
power quality data. Whether it is Northern Virginia, or Silicon Valley,
a lot of us drink from the same power pond. Its not so much a competitive
matter, as bad power will make all of us sick. Internally we all have
UPSes, generators, flywheels, etc. but we all get utility power from the
same grid.

Most of us collect a lot of data with power quality meters from our point
of view of the utility grid. But since none of us share information, none
of us really knows what the whole picture looks like. A few miles a way
the major regional power outage may show up us just a blip in the data.
And unless you were looking specifically for it, you might not have noticed.
Nevertheless, the additional data might help with the analysis, if you knew
who to ask to get it.

AOL has one of the best designed, built and operated data centers in the
world. Everyone I know who has ever seen it has been very impressed. It
would take a remarkable event to affect AOL's operation. But we are always
building based on the last earthquake, or the last hurricane or the last
power problem. So we learn every time the extraordinary happens. Other
data centers may use a very similar power design, so the extraordinary
is very much of interest.

Who knows, another data center might have already had the similar problem
and found a fix.

This is of course the extremely low-end of power quality data gathering,
but I've thrown up the scripts I use with apcupsd and rrdtool to grab the
power quality info from an APC smart-UPS in the San Diego, CA, US area.
They're at; my graphs are also up there.

If nothing else, this may be an easy way to 'webify' a subset of the data
to facilitate sharing. (note, I don't consider 120+/-5V to be much of an
issue...but the fluctuations are interesting.) They're also only as
accurate as whatever the APC uses to gather its data. But it works.

I don't have the time to support the scripts, unfortunately (mails asking
for help with them will be ignored), but they're simple enough that you
should be able to figure them out with access to the RRDtool documentation
and the README I've stuck in the package. Egregious errors I've made will,
of course, be fixed if someone tells me about them. }:>

If anyone is particularly interested I could start a link farm of other
power quality graphs such as these on this set of pages; I'm not sure how
much of that would be considered 'proprietary data', though.


On Thu, Apr 05, 2001 at 08:33:23PM -0700, Sean Donelan put this into my mailbox: