60Hz US Power Grid and precision timekeeping

I think I’m missing the point of the strategies in regulating the 60 Hz frequency of the US power grid.

I have not found one single reference that explains everything in a way that seems reasonable.

But I have pieced together an unreasonable story using many references.

Apparently, the US power grid is in about 4 segments that do not maintain phase relationships between them. I’m in the largest one, which handles the Eastern US. For mine (or for any of the others by itself), the power company recognizes that it is hard to bring generators online if you don’t know the phase you have to match.

Back in the day, people would add a generator to a polyphase AC system by stringing lightbulbs in a circle and adjusting the generator speed to make the light stop chasing around the circle, and then throw the switch. Now it’s more complicated of course.

So what they do is keep the phase under very precise control, by guaranteeing that there will be 5,184,000 cycles in any day (that’s 60 times the number of seconds in a day, 86400).

But they let this thing get several cycles out of phase during the day, and trim things at the end of the day to hit the 5184000 mark precisely at midnight.

Note that if they do it this way, it actually doesn’t make bringing a generator on line any easier. As long as you’re a big part of a cycle out of phase, the timing is worthless.

Note also that when you are bringing a generator on line to match a power grid, you obviously have access to both the generator phase and the power grid phase, so it’s not as if you need an accurate clock to do it.

I read that power companies consider the 60 Hz frequency a timing service for which they don’t charge.

I also read that on the time scale of a few milliseconds to a few seconds, the physical rotating inertia of all the generating equipment actually forms the reservoir that does load leveling, so it has to be able to speed up or slow down with some freedom. It’s obvious they don’t regulate steam turbine valves or hydroelectric gates or house-sized deisel generators on a 1 second timescale.

So - anybody an expert on this? Or can you send me someplace that is?

While we’re on the subject, everybody should go check out leapsecond.com. It’s incredible.

Wow. Thanks for this. It’s really interesting. Sorry I can’t add anything useful but your piqued my interest and I’m going off to research (well, later tonight when the work disasters are over)


It seems to me you are a bit confused about the definitions of frequency, phase, and cycles. I’ll attempt an analogy:

Imagine a circular track with race cars on it.

If all the race cars are going the same speed, they all have the same frequency…you could measure this in laps/hour.

If you designate one car as the leader, then you can measure the “phase” of all the other cars by how many degrees they are ahead of, or behind the leader. If the cars are NOT spread all around the track, but pretty much in one bunch, you could use the average positon of that group as a reference instead…and that is what is done on a power grid…the phase of the grid is the average phase of all the alternators attached to it. A few alternators might be at exactly this phase, but most will be either ahead or behind.

In order to change phase, a car has to speed up or slow down for a bit with respect to the leader, then resume driving the same speed as the leader. If it does not resume the same speed, then the phase will keep changing. This means that for part of a lap, or maybe several laps (cycles) one car can have a slightly different speed (frequency) than the leader.

You can keep track of the cycles by counting the total number of laps completed.

OK, enough analogy, lets talk about how it actually works in practice:

One thing you got completely wrong: The phase is NOT kept under any sort of control, precise or otherwise. Attempts ARE made to keep the instantanious frequency around 60 Hz. In order to do this the phase goes all over the place.

Serious effort is put into making the daily average frequency EXACTLY 60 Hz. It seems like you have a good handle on how this is done, but note that it requires tweaking the frequency above or below exactly 60 Hz to accomplish this. This is because the generating stations only have about 50% control over the frequency of the grid. The other part of the control is from the loads, which the utility does not control, and which vary constantly.

When an AC alternator is connected to a large grid, It always runs at synchronous speed determined by the frequency of the grid…

It’s intrinsic phase will lead the phase of the grid by an angle known as the power angle. If more torque is applied to the input shaft, the power angle will increase. This will of course require that the speed momentarly increase above syncronous speed. Depending on the size of the grid, and the alternator, this will VERY slightly increase the frequency of the entire grid.

When the grid frequency increases, all the motors connected to the grid speed up a little. This means they put out more power, which utilizes the extra power put in at the alternator.

Adding a motor to the grid lowers the frequency a hair, slowing down all the other motors, Thus the new motor steals power from those already connected…the generators don’t have to respond instantly. The decrease in frequency is noted at a control center, and additional capacity is brought on line to compensate.

The Separate grids were caused by the fact that the industry started out as small islands of generators and local loads. As the smaller systems grew there were advantages to interconnecting, however the larger the islands grew the more difficult it was to link them up. The US power grids are the result of physical, political, and financial barriers. The East coast and West coast development is bounded by the Rocky mountains. The Texas grid separation is a result of Texas financial and political pressure to keep Texas energy governance out of the hands of the feds (no power lines across state borders, no fed interference). At stake was not just the Texas electric business but also the Oil which could be used to produce electricity and export power out of the State. I don’t know the reason behind the cause of the NorthEastern grid.

Matching the grid frequency (phase?) to bring a generator on line is essential but not all that difficult with today’s technology. The frequency floats ever so slightly due to load but is usually within .02Hz

actually the technology today makes it much easier. Feedback systems to control generator speed can precisely match the grid frequency and its pretty much automatic.

Time correction isn’t necessarily daily. Sometimes the correction period isn’t scheduled until a error threshold is reached. Usually the largest utility in the region is responsible for coordinating the process.

I think you misunderstand something but I don’t follow your statement here. Most time correction is done during off peak times and only lasts a few minutes. During time correction ALL of the generators will slow down (or speed up) to adjust a few 100ths of a cycle to each second. The whole process could occur over a few minutes. If you were bringing a genrator on line during this time it would simply match the frequency of say 59.95hz instead of 60hz. At the end of time correction it would adjust its speed with all of the other genrators on line.


nice little market spin :smiley:

one second-no instantaneously - yes
Automatic controls will operate in order for the unit to maintain speed. The feedback mechanism for all practical purposes is instantaneous. But I see what you mean; the inertia of the rotating mass smooths out the drag on the machine but the power applied to the rotor increases or decreases instantaneously based on the turbines speed.
So - anybody an expert on this?
IANAExpert - just a DUG

This whole thread is interesting. Let me note that Quebec, like Texas is not connected to any other grid. They sell power to New England and New York, so they must have ways of synching.

Back around 1955, I was working in a lab at Penn and we actually tested the frequency occasionally. The test consisted of five tuning rods labeled from 58 to 62 and you judged frequency to about a half CPS by which ones were vibrating the most. We regularly found frequencies anywhere from 59 to 61. We were told that someone in the head offfice of the Philadelphia Electric Company simply adjusted the frequency during the night so that the clock on the wall matched WWV. I don’t think they were attached to a grid in those days. Mostly they used coal generating plants, although they owned a dam at Conowingo, just over the Mason-Dixon line in Maryland and perhaps others. But I drove over that dam frequently and it looked like small potatoes.

About modern systems being more complicated than light bulbs, and catching up at the end of the day:
"All generating equipment on a grid has to be running in sync of course, but only the smallest generators can run by just monitoring their own phase lead- that only works if the generator is small enough so that its unable to affect the phase.
"Any large generation facility has to have a reference oscillator and will actively control frequency and power output to achive the desired line voltage and frequency. A Gigiwatt is enough power output to cause large voltage and or frequency fluctuations. Even a Megawatt plant needs to regulate carefully to be sure its help rahter than hurting power quality.
"The power plants use a hierarchy of phase-locked reference oscillators. Nowadays they may be switching to GPS based references, but I suspect that since the system they have is working, it has not been changed.
"There are a few (I think its three in the US) main grid blocks, and each of them has a reference, usually at one of the main generating facilities. (I think one is at Niagra/Horseshoe Falls for the Northeast US and SE Canada, not sure.) This reference is fed (used to be by leased telephone lines, nowadays who knows?) to sub-masters in each area (approximately state by state, but that depends on generator locations, not state lines). Sub-masters feed smaller local power stations.
"Again, these are all phase locked, and they do something to compensate for propagation delays when they set a power plant’s reference up.
"So they have a reference. But the original question was, what do they do with it- how good is the frequency regulation? The engineer I talked to (and this was a while ago) said that frequency does vary with load. The grid managers do try to run fast (and therefore high line voltage, by the way) when they are lightly loaded at night to “catch up” to 5184000 cycles per day, they do try to regulate day by day when possible. Unless they go offline (blackout) when all bets are off.

Another reference to cycles per day matching at midnight (though worded vaguely enough that I don’t know midnight is to be taken literally) from
“An excess of generation over load will cause the frequency to raise, a deficit of generation will cause the frequency to fall. The power system is operated to provide exactly 5,184,000 cycles per day, midnight-to-midnight and the power output is adjusted as necessary to cause the frequency and cycle count to drift back to where it should be.”

A reference with numbers who doesn’t think cycles per day is controlled
"The frequency variation in the American power grid on any given day is between 59.971 and 60.035 HZ.
"Frequency varies with grid load because a higher load will load the generator and slow it down ever so slightly.
“I really don’t think that a set number of cycles is maintained each day.”

Unattributed article quote:
“The entire North American grid operates at a nominal 60 Hz on four big circuits (one per interconnection) linked by high-voltage dc tie lines. As part of this grid, PJM also must maintain 60 Hz. If a load suddenly draws more current, then a drop in voltage and frequency will occur across the grid, affecting the speed (frequency) of the generators supplying current to the load.”

Unattributed article quote:
"The line frequency does vary slightly over the course of the day. (usually decreasing during times of heavy load). At the end of the day, they run slightly faster or slower as needed so that the total line cycles for the day is EXACTLY 5184000. That is, they correct so that the number of line cycles is correct in each midnight-to-midnight period.
“Why is this done? I mean why does the power line need to be EXACTLY 60Hz? Most generators are synchronous machines, when they are connected to the grid, you put torque in and they put power into the line. If you load them down, they are motors and draw power from the line. When you start them up, if the rotation speed and phase is not exactly in synch with the grid, they draw huge fault currents until they synchronize or blow the overload breakers open. So they HAVE to be able to synchronize.”
OK, then. This last quote really catches it. BubbaDog, when you say you think I misunderstand something, I agree, and it’s right here. They are making the point that hitting the right cumulative cycle count every midnight is somehow helpful whenever they need to bring a generator online, and that this is more or less at the heart of the frequency regulation strategy. I don’t think that makes any sense. Sure, you need to adjust the phase relationship between generator and grid so they match before you connect them, but this is only meaningful right around the time you do it. What happened last midnight or will happen next midnight is irrelevant. And yet they appear to be going to a great deal of trouble to achieve this capability, with atomic clocks and leased phone lines and what all.

This “exact count at midnight” thing seems to turn up in many places. Maybe it’s some electrical engineering meme that got spread through misunderstanding. It could be that big grids and generators with large moment of inertia have grown more difficult to work with. I know some of the big regional blackouts over recent decades have been blamed on subtle system dynamics. That’s what I was thinking when I described modern systems as being more complicated. I think the first quote above hints at this.

It keeps coming and coming:

A very pleasant if elementary discussion of time from NIST mentions the grid at

“The electric power line is a very good frequency source for many applications. A clock driven from the power line will keep excellent time. This is because the power system is carefully controlled to maintain its frequency within definite limits. Each power company is notified in advance to set its frequency to a particular value, so that the millions of clocks on the system (in homes and offices across the country) will gain or lose time as required to keep the clocks correct. The time corrections are usually done at night.”

Fascinating presentation about NIST involvement with phasor stabilization of the grid for reliability improvement
Calibrations Show that System Meets our Goals
Less than 0.01 % magnitude error
Less than 0.2 µs time error
< 4.4 mdeg phase error at 60 Hz

Authoritative article from NIST mentions among other things a 10 ms requirement for generator phase control in the grid (in their Table 3)
"The electric power system in North America consists of many subsystems that interconnect into several massive grids that span the continent. The system delivers the 60 Hz AC frequency to many millions of customers by matching power generation levels to transmission capability and load patterns. The entire power system relies on time synchronization, and synchronization problems can lead to catastrophic failures. For example, the massive August 2003 blackout in the eastern regions of the United States and Canada was at least partially caused by synchronization failures.
"The timing requirements of the power industry vary (Table 3), because different parts of the system were designed at different times, and the entire system has evolved over many years. The older parts of the system have less stringent timing requirements because they were designed using technologies that predated the Global Positioning System (GPS). The newer parts of the system rely on the ability of GPS to provide precise time synchronization over a large geographic area.
"Since electrical energy must be used as it is generated, generation must be constantly balanced with load, and the alternating current produced by a generator must be kept in approximate phase with every other generator. Generation control requires time synchronization of about 10 ms. Synchronization to about 1 ms is required by event and fault recorders that supply information used to correct problems in the grid and improve operation. Stability control schemes prevent unnecessary generator shutdown, loss of load, and separation of the power grid. They require synchronization to about 46 µs (±1° phase angle at 60 Hz), and networked controls have requirements one order of magnitude lower, or to 4.6 µs (±0.1° phase angle at 60 Hz). Traveling wave fault locators find faults in the power grid by timing waveforms that travel down power lines at velocities near the speed of light. Because the high voltage towers are spaced about 300 meters apart, the timing requirement is 1 µs, or the period of a 300 meter wavelength. Newer measurement techniques, such as synchronized phasor measurements, require time synchronization to Coordinated Universal Time (UTC) to within 1 µs, which corresponds to a phase angle accuracy of 0.022 ° for a 60 Hz system. A local time reference must be applied to each phasor measurement unit, and GPS is currently the only system that can meet the requirements of synchrophasor measurements. Commercial phasor measurement units that receive GPS signals are shown in Fig. 4.
“The 60 Hz frequency delivered to consumers is sometimes used as the resonator for low priced electric clocks and timers that lack quartz oscillators. The legally allowable tolerance for the 60 Hz frequency is only ±0.02 Hz, or 0.033 %, but under normal operating conditions the actual tolerance is much tighter.”

Also, BubbaDog, I keep rereading the quote ending in “So they HAVE to be able to synchronize”. Maybe I misread it. Maybe he stresses the 60 Hz is for synchronizing generators, not the counts-at-midnight thing. But still it keeps appearing out there - why?

And another thing. This gets at the whole system dynamics thing.

When you run a synchronous motor or generator on a system, it can oscillate ahead and behind the phase of the system. You might then worry about whether those oscillations tend to damp out, or tend to grow to the point of jumping a cycle and perhaps catastrophic failure. If I remember right from some work I did a few years ago with synchronous motors, this is a real problem, because most of the things you do to make motors nice and efficient also raise the Q of this harmonic system and leave them vulnerable to oscillations growing uncontrollably. The same thing happens in stepper motors, which are synchronous motors having PM cores and two phase (90 °) windings, and which are often run in open loops without feedback. They actually make viscous oil dampers for stepper motors, cans that mount on the shaft and float an annulus inside a container of viscous oil. They budge the motor off of high Q and towards a rapidly exponentially decaying oscillation, shifting it to the left in the laPlace s-plane. But in big synchronous motors this is very unweildy and messy. They manage it, I think, with things like killer windings that short out the leading and trailing edges of the salient poles so that variation of the magnetic field strength there generates heat, making the oscillations die down.

Sounds like the 5184000 cycles per day is a generation/billing issue that has little relevance to the end user.

60Hz was probably chosen intitially for convenience in the control devices available at the time (high precision gearing made for clocks and timers have an affinity for 60) and it’s high divisibility means multiphase current peaks are likely to have an integer degree of separation… making generator design and testing a little easier.

Once you’ve got a lot of electrical equipment out there expecting a given frequency on the line, you’re pretty much stuck with it.

If you have multiple independently regulated generators online, it’s important to maintain constant frequency or else it’s very difficult to control loading of any individual machine. Raising controlled frequency of an AC machine increases carried load (NOT frequency)… decreasing controlled frequency unloads the machine (frequency, again, does not change significantly). Unexpected/uncontrolled changes in machine loading can be unpleasant.

I used to have a digital clock, built from a kit at a time when digital clocks were unusual. It used a secondary tap from the power transformer to produce a 60 Hz square wave. This signal drove the counter/divider logic in the clock. It was interesting to compare it to WWV. It was never off by more than a second or two. While its short-term accuracy was mediocre, its long-term accuracy was excellent, thanks to the people who run the grid.

Many electro-mechanical clocks used AC synchronous motors, which made their accuracy dependent on the power grid’s frequency regulation. Many older computers also used the line frequency to generate their real-time clock. Each cycle on the power line would generate a clock interrupt. The computer’s operating system would count the clock interrupts and use them to update the system’s time and date.

Have heard rumors for years that the Xcel Energy Shoshone Power Plant (hydro-electric) in Glenwood Canyon, CO is the 60hz ‘clock’ that the Western Power Grid is timed. Meaning that it, Shoshone, is the standard for all the powerplants to sync to.

Any truth to this?

What happens if all the plants on the grid go down? They cannot all just fire up and run power to the grid or, I suppose, a huge physical timing sequence of plants coming on-line in a certain order would be required.

I understand about the atomic clock with the GPS, but it would make sense for a physical plant to be used for phasing and frequency and such. A hydro plant makes most sense as they are rarely off-line and are fairly ‘hands-free’.

Someone will provide an answer in a jiffy.

So 4 1/2 years on, does anyone have a verifiable reason why they run the frequency slightly fast at night, to “catch up” with the cycles lost during the day, instead of just running at the correct frequency?

Is there an answer to which of these came first?
a) electric clocks rely on the power supply keeping up, which it was already doing for whatever reason
b) the power supply makes up the cycles lost during the day, to maintain the accuracy of electric clocks.

No doubt a) came first. The first electric clocks must have relied on the AC cycle to maintain constant speed, but would have required periodic adjustment. Then, if it is true, the electricity providers would have attempted to keep them accurate by adding/subtracting cycles at night.

You say a) came first, but then your description makes it sound like you think b) came first.

I’m saying the cycles would have been the only practical way to maintain relatively good time with the first AC clocks. It wouldn’t have been completely accurate based on variations in the cycle, so eventually, at some later date, the compensation would have been introduced so that electric clocks didn’t need to be reset on a regular basis.

As a (very) side issue to this thread, there’s a longstanding and very bitter dispute going on between the provinces of Quebec and Newfoundland & Labrador about Quebec’s sale of power generated at the Churchill Falls plant in Labrador.

Newfoundlanders feel, very strongly, that they’re being robbed to the tune of billions of dollars by Quebec.

Just checked an AC outlet in my garage with my trusty Fluke meter. 58.73 Hz. Maybe that’s why my electric clocks lose a bit of time each month.

periodically measure it and see if it varies.

IANAPE (Powergrid Engineer), but that seems unreasonably low to me. Are you on a local generator or something, instead of one of the main grids?