This is something I've been curious about for a long time. Let's say you're a utility trying to connect two segments of a distribution grid. Both segments will have about the same frequency, because they are both connected to the transmission grid, but the phase angles of their voltages aren't necessarily related to each other. I always assumed the procedure involved adjusting generators on one or both sides to bring those phase angles as near alignment as possible and then closing the switch.

A common AC distribution voltage in the U.S. is 7.2kV line-to-ground, though. That seems pretty unforgiving when you're trying to close a switch. If your synchronization procedure is off by even a tiny amount, you're going to see massive amounts of current flow.

Just to put some back-of-the-envelope numbers to it, let's say your synchronizing procedure is accurate to about 0.5%; maybe you're closing slightly out of phase or there is some small error in your sensors. That's still 36V across the switch, which we'll say has an impedance of 0.1 ohms. Now suddenly we have 360A at 7.2kV for 2.5MW. I know that's a transient current and both sides will get yanked into voltage agreement with each other, but that still seems like a lot! And that's not to mention that some networks are run at much higher voltages than 7.2kV. Is this just something utility switching equipment is built to handle, or are my estimated numbers off?

  • 1
    \$\begingroup\$ So far as I remember, switches for high-power applications are typically called contactors. This can help guide your research. \$\endgroup\$
    – Reinderien
    Dec 1, 2018 at 5:47
  • 1
    \$\begingroup\$ Here is how the procedure was described to me (for lower voltage generators) . When the frequencies are close, the generators go in and out of phase very slowly. If you put a light-bulb (or a voltmeter, I guess) across the two generators, you can observe the voltage. When it hits zero, you close the switch and connect the generators. After that, they will remain in sync by themselves (if one tries to overspeed, massive mechanical torque arises). \$\endgroup\$
    – user57037
    Dec 1, 2018 at 7:36
  • \$\begingroup\$ Thanks Reinderien and mkeith for your responses! I've seen the lightbulb method for synchronizing smaller generators at lower voltage. The basic single-generator mechanism of rotor speed to increased electrical generation to mechanical torque makes sense to me, I was just always kind of skeptical that a similar process could translate well from ~240V to power-grid level voltages. \$\endgroup\$
    – kb4444
    Dec 1, 2018 at 21:45
  • 1
    \$\begingroup\$ 36V across 0.1 ohms is not 2.5MW. Your math is incorrect. \$\endgroup\$ Dec 2, 2018 at 7:38
  • \$\begingroup\$ You're absolutely right, thanks - that was fuzzy thinking on my part. I had just multiplied the current by distribution voltage level, but of course there would be transformers involved in getting it up to 7.2kV. That clears up all my confusion. \$\endgroup\$
    – kb4444
    Dec 2, 2018 at 23:19

1 Answer 1


Firstly, the process (simplified): -

enter image description here

Picture taken from this very good article and I would recommend reading it for more detail. It is a simple line diagram and there are components not shown that will alleviate the connection should any small amount of non-synchronous voltage be present.

I'm no expert on this but I would consider the following to be relevant to achieve synchronization with as little disturbance as possible: -

  • The timing error might be 0.5% (as per the original question) so to minimize the disturbance, "connecting" would (or should) occur close to the peaks of the sinewaves of the two voltages.
  • 0.5% timing error is a phase error of 1.8 degrees and that could produce a voltage error of \$\sqrt2\$ x 7200 x (Sin (90) - Sin(88.2 deg)) = 5.02 volts.
  • This of course relies on switching happening near the peak of the sinewave and the switching/connection device to be quick (or compensated for)
  • Reactors (inductors) placed in series with the connection will buffer the connection transient and quite possibly these can be bypassed pretty soon after a connection is verified.

I'm not expert so maybe someone else can shed more detail.

  • \$\begingroup\$ Much appreciated! I'll take a look at your article as soon as I have a chance. Switching near the peak of a sinewave and placing series inductors are both good ideas. \$\endgroup\$
    – kb4444
    Dec 1, 2018 at 21:48
  • \$\begingroup\$ (hit enter too soon) … Your figure is a helpful one for refining my question, too. My confusion really centers on those potential transformers. They would affect both the voltage magnitude and the phase angle of incoming sine waves, which would translate into some voltage mismatch even if the sync lights worked perfectly. Then it wouldn't take much voltage mismatch across a very low-impedance switch to turn into a lot of current. \$\endgroup\$
    – kb4444
    Dec 1, 2018 at 21:58
  • 1
    \$\begingroup\$ The transformers have to be matched of course. Think calibration / setup procedures involving accurate test equipment and possibly using taps on the transformers per turn to match the outputs to very accurate levels. It’s important enough to spend literally an hour setting up these thing. \$\endgroup\$
    – Andy aka
    Dec 1, 2018 at 22:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.