Multiple generators in phase. How?

[Harmonious Discord]

How do power companies combine multiple generators for power output and keep each in proper phase.

[Public Animal No. 9]

How do power companies combine multiple generators for power output and keep each in proper phase.

It's relatively simple to do this. The operator brings the generator up to speed by adjusting the turbine or diesel generator. There is a meter that shows the difference in phase between the individual generator and the grid. By slightly speeding up or slowing down, the single generator phase can be synchronized with the grid, and the breaker between the generator output and the grid is closed. Once that happens, the operator can then start adding steam (or fuel to a gas turbine) to begin loading the generator and sending more power on to the grid.

Operators usually try to synchronize as they are speeding up so that there's a little extra power going to the turbine when the unit synchronizes. They do that so that the grid isn't running the generator as a motor.

When the system (grid) load increases, the frequency starts to drop a little, and each generator that's connected automatically senses the drop in speed and adds a bit more power to maintain the 60-cycle (or 50 in Europe) frequency.

[Chronos]

And once the generators are running in phase, no special effort is required to keep them in phase. In fact, it would take a Herculean effort to get them out of phase, since a generator which started to fall behind would begin acting more as a motor, and speed back up to stay in phase.

[David Simmons]

Here is a pretty good basic introduction to power grids. (http://www.av8n.com/physics/power-grid.htm#htoc6)

Now, I have a question. The long term average frequency must be pretty close to right on because electric clocks keep good time. How do they manage that?

When the load increases the frequency drops a little before it is brought back up. When the load decreases the frequency rises a little before it is brought back down. Chances are these won't exactly cancel so what do they do, count cycles over a day and adjust frequency as needed?

[Public Animal No. 9]

...

When the load increases the frequency drops a little before it is brought back up. When the load decreases the frequency rises a little before it is brought back down. Chances are these won't exactly cancel so what do they do, count cycles over a day and adjust frequency as needed?

When the frequency "drops a little," it's very little - far less than one cycle per second. When frequency gets even as low as 59 Hz, that means there is a huge drag on the system - a lot more load than there is power available - and generators start tripping off line to protect themselves. That reduces the power even more, which is one way to end up with a major blackout.

In short, it would take quite a while for any noticeable change to impact clocks. Over the course of a typical day, the load goes up and down quite a bit, so there is ample opportunity for the changes to cancel out. In any case, the link you provided tells (see sect. 2.6) how any adjustments are made to ensure that clocks are kept relatively accurately.

However, Friday afternoons tend to be when clocks slow down the most, so maybe that's due to too many people heading home and turning on the AC and opening the refrigerator. ;)

[David Simmons]

When the frequency "drops a little," it's very little - far less than one cycle per second. When frequency gets even as low as 59 Hz, that means there is a huge drag on the system - a lot more load than there is power available - and generators start tripping off line to protect themselves. That reduces the power even more, which is one way to end up with a major blackout.

In short, it would take quite a while for any noticeable change to impact clocks. Over the course of a typical day, the load goes up and down quite a bit, so there is ample opportunity for the changes to cancel out. In any case, the link you provided tells (see sect. 2.6) how any adjustments are made to ensure that clocks are kept relatively accurately.

However, Friday afternoons tend to be when clocks slow down the most, so maybe that's due to too many people heading home and turning on the AC and opening the refrigerator. ;)So, over the course of a year the accidental ups and downs equalize to the point that they don't accumulate a time error?

Or maybe they coun't on a occasional power outage which results in a resetting of clocks. ;)

[mks57]

So, over the course of a year the accidental ups and downs equalize to the point that they don't accumulate a time error?

Or maybe they coun't on a occasional power outage which results in a resetting of clocks. ;)

I used to have a Heathkit digital clock that used the power line frequency as its frequency reference. It was never in error by more than a few seconds. The power company would correct any accumulated errors by running the grid a little fast or slow during the wee hours of the night. It was a good example of a frequency reference with mediocre short-term stability and excellent long-term stability.

Once you get off a well-run grid, all bets are off. In some places, frequency regulation is very poor. I've seen this when playing magnetic tapes that were recorded in certain countries. The original tape recorders used AC synchronous motors for speed regulation, which resulted in significant frequency shifts when the tapes were later played back at the correct speed.

[Rysto]

It's relatively simple to do this. The operator brings the generator up to speed by adjusting the turbine or diesel generator. There is a meter that shows the difference in phase between the individual generator and the grid. By slightly speeding up or slowing down, the single generator phase can be synchronized with the grid, and the breaker between the generator output and the grid is closed. Once that happens, the operator can then start adding steam (or fuel to a gas turbine) to begin loading the generator and sending more power on to the grid.
Surely this isn't done by hand. Don't they have a computer that can do this?

[jayjay]

Now I have Carmen Miranda's Ghost stuck in my head...

Carmen Miranda's ghost is haunting Space Station Three.
Half the staff has seen her, plus the Portmaster and me.
And if you think we've had too much of Cookie's homemade rum,
Just tell me where those basket-hats of fruit keep coming from.

...

We don't know why we're haunted here, or why it's her that haunts.
We've got a betting-pool for all who wonder what she wants.
The best odds say she likes the rhythm of the station's drive;
They didn't have phase-generators while she was alive.

[David Simmons]

Surely this isn't done by hand. Don't they have a computer that can do this?This paper from 1979 (http://www.freepatentsonline.com/4249088.html) shows that automatic devices have been around for quite a while.

The paper references prior patents from 1974 and 1975 for different means of doing the same thing.

I'm not a power grid specialist but I would expect by now that computer controlled digital systems would be all the rage.

[Triskadecamus]

I'm pretty sure the cycles are counted. At the end of a predicted time period, the total is compared to the desired total, and adjusted, using heterodyning to add or subtract the precise number of cycles at the next predicted interval. A sixtieth of a second is not going to matter much, for most applications, but in fact the average is going to be much more accurate than that.

Tris

[engineer_comp_geek]

Surely this isn't done by hand. Don't they have a computer that can do this?

It's almost as easy as throwing a light switch. There are two ways of syncing a generator to the grid. One uses an instrument called a synchroscope, which is a fairly simple meter. You just wait until the meter registers a zero phase difference and throw the switch. Once you get the generator close to line frequency, the phase difference changes fairly slowly, so hitting it at exactly the zero point isn't difficult at all. The second method involves connecting 3 lights between each of the phases. When the generator is out of phase with the grid, the lights turn on. When they are in phase there's no voltage difference, so the lights go out. You watch the lights for a bit to get the pattern, and throw the switch when they are at the mid point of their off cycle.

Imagine a light bulb turning on and off about every ten seconds. You have to throw a switch while it's off. It's that simple.

Of course, if you throw the switch at the wrong time the generator tends to come apart in a rather spectacular fashion, which can be a very exciting event when your generator is the size of a small house. On big expensive generators they do use automatic equipment to make sure the generator is in phase before it connects it to the grid. Smaller and older generating systems will likely not have any automatic equipment to protect the generator.

Once the generator is attached to the "grid" it's about impossible to get it out of phase. If you try to speed it up, it generates more power to the grid. If you try to slow it down, it generates less power to the grid. Try to slow it down enough and it will actually suck power from the grid and become a motor instead of a generator.

It used to be that the power companies were required by law (in the US at least) to keep the long term accuracy of the grid to within 1 second. I'm not sure if they've changed the law now that synchronous motor clocks have all but disappeared.

[si_blakely]

Phase synchronization is one of the barriers to in-home co-generation units - if you are selling power back to the grid, it has to be in phase and synchronized correctly. Small sized units probably try to drift a little (due to low mass) but also correct easier.

I remember (a few years ago now) of a project run by a polytechnic in NZ to build power units using Truck turbochargers and alternators (powered by natural gas) - 10kW electricity and 10kW heat output. It was the electronics around the mechanical equipment that was the tricky bit. Only now, here in the UK, can I buy a co-generating boiler system for my house, with slightly lower power specs.

Si

[Carson O'Genic]

Public Animal No.9,
Thanks for your postings here.Explanatory even to this box o' rocks.

You others,carry on.Sorta expect that from you.

[Harmonious Discord]

Thanks for the input. I hadn't considered the motor aspects of the generator, and the power of the others feeding it if out of phase. Like was said, with generators as big as they are, you could have a spectacular failure if brought online incorrectly.

I'm very happy that two posters are new, and one is brand new. Welcome aboard.

[billyb0b]

Harmonius Discord did you mean phase or frequency? Seems the ost of the answer here dealt with frequency (failrly easy to regulate) rather than phase. Someone did mention phase sequencing lights, but didtn't explain how they work. I used to work on mid-sized generators for the Air Force and used those phase sequencing lights/meters often, but never knew how they worked. Anyone know?

[Squink]

If the frequencies of the generators are slightly off, their phase relation will drift, and you'll see the resulting beat frequency (http://en.wikipedia.org/wiki/Beat_(acoustics)) in the phase sequencing light.

[Hari Seldon]

Here is a pretty good basic introduction to power grids. (http://www.av8n.com/physics/power-grid.htm#htoc6)

Now, I have a question. The long term average frequency must be pretty close to right on because electric clocks keep good time. How do they manage that?

When the load increases the frequency drops a little before it is brought back up. When the load decreases the frequency rises a little before it is brought back down. Chances are these won't exactly cancel so what do they do, count cycles over a day and adjust frequency as needed?

Back around 1955, I had a friend who had had a tour of a power station. Of course, back then they didn't have computer controls and they didn't have giant power grids either. We had a frequency meter (essentially a bunch of tuning forks tuned to 57, 56, ..., 63 cycles/second (we didn't call it herz then) and we could see that the power varied from less than 59 to more than 61. But the power station had an accurate crystal controlled clock and an ordinary electric clock. In the middle of the night, they varied the frequency as necessary to bring the two clocks into sync. Of course, after a power failure or at change to DST, they made no attempt to get the clocks in phase, except by resetting the ordinary electric one.

Presumably what they do today is a high tech version of the same thing. I wonder, do they adjust for leap seconds?

[David Simmons]

Like was said, with generators as big as they are, you could have a spectacular failure if brought online incorrectly.With all of the generators being magnetically coupled together you can think of them as one huge generator with a colossal inertia. A single out of phase generator, no matter how big, will be jerked around pretty good when connected.

[Harmonious Discord]

I meant phase.
The frequency has to be the same to maintain the proper phase.

[robby]

When I was in the Navy on a submarine we routinely connected each of our two ship's service turbogenerators (SSTGs) to each other and to shore power.

There was a synchroscope on the electrical panel. You set it up so that you could see the frequency difference between the the two sources being connected.

The synchroscope consisted of a meter with a pointer that rotated in a clockwise or counter-clockwise direction. The clockwise direction was labeled "FAST" and the counterclockwise direction was labeled "SLOW." The rate of rotation corresponded to how much the two sources differed in their frequencies. The direction of rotation corresponded to which source was higher in frequency.

You were supposed to close the breaker that synchonized the two sources when the meter was going "slowly in the FAST direction." This meant that the SSTG was slightly higher in frequency than shore power, so it would pick up load and not act as a motor, and the fact that the pointer was moving slowly meant that SSTG was only slightly higher in frequency than shore power.

The breaker needed to be shut when the two sources were exactly in phase. The pointer would be at the 12 o'clock position at this point. In practice, with the pointer moving slowly in a clockwise direction, the breaker was actuated at about the 10 o'clock position so that the breaker actually shut at the 12 o'clock point.

I was an officer, so I only watched all of this. (I did do it myself a few times in training, though.) I had electricians who could switch from one power lineup to another, all manually, in seconds. One guy could synchronize and parallel two power sources before the synchroscope meter even made it once around.

[LSLGuy]

Multi-engine aircraft aren't much different from submarines in this regard.

Older multi-engine jets, say airliners from the 707 through the early 737 and 747 used a similar system. Each engine drove an AC generator through a hydraulic pump/motor assembly called a "constant speed drive" or CSD. The CSD tried to maintain a constant output (ie generator input) shaft RPM regardless of engine RPM. That way the generator's frequency would also be constant, at least within the tolerance of the CSD's governor.

Each generator was connected to a bus which carried some fraction of the total aircraft load. As well, there was a "sync bus" which connected all the generators together. This enabled the entire system to weather a generator dropping off line with no interruptions in supply to anything as long as total load was below the total output capacity of the surviving generators.

It also meant all the generators had to operate on a common frequency and exactly in phase. After each engine was started there was an elaborate ritual the flight engineer (3rd crewman in the cockpit) had to go through to get the newly started engine's generator synced with the others before connecting it to the sync bus & the loads.

Each unit had a frequency meter and phase difference lights. You'd adjust the free generator's freq via a knob while monitoring the meter until it matched the indicated freq of the sync bus, then turn on the phase difference lights & fine-tune the free freq until you had a very slow (1 cycle / 10 sec) beat. Fnally, you'd close the switch at the center of the off part of the blink cycle.

Given that CSDs were mechanical beasts whose behavior depended on temperature, difficulties often arose on frigid mornings & sometimes folks had to wait a few extra minutes for the hydraulics to come up to temp before the situation would stabilize enough to connect sucessfully. Many times there was a little breath-holding when the switch was thrown, since you were holding up traffic & the Captain wanted to get under way NOW. This had to be done for each engine start after the first, so a 707 or 747 engineer had 3 chances to screw up on every flight.

There was a mechanical weak link in the generator input shaft and if you really hosed up & tried to connect when grossly out of sync the link would break. This prevented catastrophic generator damage, which was good, but it also meant the flight was cancelled until it could be repaired and you got to go talk to the boss about your ineptitude. And who knew how many times that link had almost been broken from previous ham-handed connects? So most of us treated those links & connect operations with kid gloves.


More modern designs simply keep each generators' loads isolated in normal ops & if a generator fails an automatic relay will connect the now-unpowered bus to one of the others which still has a generator. As long as the different busses are generally near correct voltage & freq, ie.. +/- 10%, it'll work fine. The loads see a momentary (<1/10th sec) loss of power, but as long as they're designed for that possiblity, no problem. No more precise syncing is required.

There are also some uber-modern designs that do away with the CSD altogether & produce AC power of widely varying frequency, say 200 Hz at idle & 450 Hz at full power (400 Hz being the historical standard on aircraft). It requires a little different design for some of the loads, but there was never a strong reliance on exact Hz-itude anyhow, so the changes are comparatively minor.

[engineer_comp_geek]

Someone did mention phase sequencing lights, but didtn't explain how they work. I used to work on mid-sized generators for the Air Force and used those phase sequencing lights/meters often, but never knew how they worked. Anyone know?

It's easy to explain with pictures, not so easy to explain in words.

Connect a light between the output of a generator and the "grid" (or another generator). The generator is going to output a voltage sine wave, and the grid is going to have a voltage sine wave. If the two are perfectly in sync, then there won't be any voltage difference between them at any time during the sine wave, so the light will not light. If they aren't in sync, then there will be a voltage difference and the light will come on. The farther they are out of sync, the brighter the light will be.

If the generator and the grid aren't even close in frequency, then the lights will flash so fast as to be meaningless. But, if you get the generator and the grid very close in frequency, the lights will flash much more slowly as the sine waves move close to phase then farther away from being in phase. The closer the generator and grid are in frequency, the slower the lights will flash.

Three phase generators have three such lights, one for each phase. When you watch them, all three lights will get brighter then dimmer then brighter then dimmer. When the lights are at their dimmest (i.e. OFF) the three phases are in sync and it's safe to throw the switch.

If all three lights don't get bright and dim at the same time, chances are you have two of the phases reversed, which is a very bad thing. If you like having your generator all in one piece, don't throw the switch until you get the wiring straightened out.

[BubbaDog]

So, over the course of a year the accidental ups and downs equalize to the point that they don't accumulate a time error?

Or maybe they coun't on a occasional power outage which results in a resetting of clocks. ;)

The National Electric Reliability Council (NERC) has rules for time correction.

They very closely monitor the time in relation to power grid frequency.

They just count the cycles and compare it against NIST time. When the error becomes too great (10 sec) the local coordinating council will notify generating entities to shift the frequency by plus or minus .02 hz at a prescribed time (usually early am).

You can read can read an explanation by one of the coordinating councils on page 26 of this. (http://www.pjm.com/committees/mrc/downloads/20070109-item-02-manual-m12-dispatch-operations.pdf)