[time-nuts] Features of a Precision Clock?

Bill Hawkins bill at iaxs.net
Sat Oct 7 16:20:48 EDT 2006

Tom Van Baak wrote,

"If it wouldn't be too much to ask, yes, I think several
of us would be interested in how, and under what rules,
they tweak the frequency."

It helps to have a mechanical understanding of the problem.
By the time you get to the end of this, you should know
why the frequency can't be continuously regulated.

If you deliver steam to a turbine from a large boiler, the
speed of the turbine is limited by the rate of steam flow
and the pressure drop across the turbine. Flow times drop
gives you the power from the boiler to the turbine, just
like amps times volts.

A throttling valve is put in the steam line to regulate the
power to the turbine. The turbine will rotate at a speed
determined by the power balance between the steam and the
shaft load. Increase the load and the turbine slows down.

The load is a synchronous generator. This doesn't mean much
for an isolated turbine and generator combination. If you
connect thousands of generators to a common distribution
system, they are all locked together in speed, but not phase
angle. As Dean said, the phase angle varies with the power
flow. The frequency varies with the power balance.

To emphasize the common speed aspect of the power system,
if one generator is disconnected from the system (perhaps
because its hydrogen cooling gas has developed a leak) then
it must be re-connected very carefully. The steam throttle
is adjusted to bring the speed of the generator (and its
frequency) up to match the distribution network frequency.
Then a phase indicator is used to detect the phase angle
difference. The breaker connecting the generator to the
power system can be closed when the phase angle is zero.
If the angle is not zero, the power of thousands of generators
tries to bring the one generator into phase alignment instantly.
The mechanical forces can do great damage before the breaker

So here we have this network of thousands of turbines and
generators all locked to the same frequency because they are
synchronous machines. In effect, we are back to the simple
case of the speed being determined by the power balance
between steam and electrical load. But we really have a network,
and the ability to measure power flows at various points in the

Electric utilities have joint power dispatching centers. The
dispatchers work to balance the power in the system by telling
power generation stations how much power to generate. This
works in large scale systems. You can't tell the users how
much to consume, you just try to meet the demand.

Small generating plants and large networks maintain a clock
attached to the power line and another from a standard, like
WWV before GPS. The station manager adjusts the fuel flow to
the boilers to alter the power balance to change the turbine
speed. The throttling valve can't be adjusted with sufficient
precision to do frequency control. Large or small, the power
line clock loses time when there isn't quite enough fuel and
gains when there's a bit too much.

Large networks barely stay ahead of demand with new power plants.
There are "peaking" plants that only generate when the demand
exceeds the base capability of the boilers. They use expensive
fuel for rapid response (gas turbines) but they hold the power

If you measure the difference between power line time and GPS
time over 24 hours, you will see the line clock lose time
during the day, when loads are highest, and regain that lost
time in the very early morning, when loads are low. The
dispatchers tell generating stations to add a bit more fuel
to bring the line clock back to match GPS or a little ahead,
anticipating the day's demand.

Why not use automatic frequency control? Because the controllers
in the different stations would fight each other, because they
are tied together by the synchronous network. In other words,
it is not possible to regulate a stable speed/frequency with
more than one integrating controller - and one controller can't
regulate enough power to stabilize the network - unless the
controller is a dispatcher able to tell many stations what to do.

More than I intended to say, but I hope someone finds it useful.
None of the above is guesswork. I've been interested in this
since a summer job on an island in 1955. I've not talked to a
dispatcher, but I have confirmed the details with a control
engineer at a large power station.

Bill Hawkins

More information about the time-nuts mailing list