[time-nuts] 60hz disciplined watch

[Jim Lux]

On 4/20/11 9:39 AM, Poul-Henning Kamp wrote:
> In message<4DAF0B0D.4080501 at earthlink.net>, Jim Lux writes:
>
>> I found this in a generator interconnection agreement:
>>
>> This frequency response control shall, when enabled at the direction of
>> CAISO, continuously monitor the system frequency and automatically
>> reduce the real power output of the Asynchronous Generating Facility
>> with a droop equal to a one-hundred (100) percent decrease in plant
>> output for a five (5) percent rise in frequency (five (5) percent droop)
>> above an intentional dead band of 0.036 Hz
> Neither that text nor any other I have been able to find, guarantee
> that the integral (= long term average) of the frequency will
> converge on 60Hz, only that the instantaneous frequency will stay
> in an particular range.

There's the "integrated time shall not deviate more than 2 seconds from 
actual" sort of requirement.  I don't know what the integration interval 
on that is, but I think it implies that in any given day, there will be 
60*86400 cycles plus minus 120 cycles. And that further, a sequence of 
+120, +120, +120 on successive days wouldn't be allowed.

I also noted that they seem to want the "corrections" to be done at the 
top of the hour (they speed up or slow down the 60 Hz until they're 
realigned)..  I don't know, off hand, how much difference that would 
make in power flow (say they adjust by 0.01 Hz... over 1 second, that's 
a phase shift of 3.6 degrees, which is pretty big in the power 
management world)


>> tvb's data at http://leapsecond.com/pages/mains/  [...]
> That data confirms what I said:  Maybe if you average over a couple
> of months, but the practical problems related to doing that are
> enormous.
>