Generator operating frequencies. 27

[HamburgerHelper]

I have thought about this for awhile and I don't understand it. So, let's say you have generation spread out over a large region and there is a disturbance in the system that causes one of the generators to be at a lower or high frequency than the rest of the generators in the system and the controls don't work to bring that generator back up to normal frequency. What happens? I have a hard time understanding this because in my mind if a generator is operating at a different frequency than the rest of the system, that generator or island around the generator is effectively isolated from the rest of the system from a power flow perspective. The rest of the grid is going to try to motor or add generation to it as the phase angle of the different frequency generation slips around the rest of the grid. I just have a hard time grasping why a generator can operate for example at 59 hz while the rest of the grid is humming along at 60 hz.

 

[ScottyUK]

Hey crshears, I'm a generation guy. Only our view of the world counts.

I guess UFLS is set to operate slightly before the generators start tripping off to protect themselves. Once the generators start falling away there will be a domino effect as the frequency falls further with each machine trip, leading to system collapse.

 

[crshears]

I used to know what you mean about generation being the only thing that counts...

First I was a steam generation guy, then transmission & distribution, then small hydraulic generation, then more T & D, then really big hydraulic generation, and finally T & D again since 1997.

So, being Canadian and with apologies to Joni Mitchell, I can say, "I've looked at grids from both sides now..."

CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[crshears]

MBrooke wrote: I think this is possible if the condition is spread across a very wide range... Ie Quebec say 60Hz, Ontario at 59.87, NY at 59.81, NJ at 59.76, Maryland at 59.65, North Carolina at 59.61, ect ect with the frequency dipping down to 59.00Hz at say Mississippi. Power would flow North to South and every generator would be turning slightly slower than the one above it. Not enough to motor or skip poles, but just enough to be lagging physically.

I had to think long and hard about how a system could be be made to function in this way; try this on for size...

Conceive of each of the system pockets mentioned in the quote above as interfacing with the next one by way of a very high quality induction coupling where transfer of power across said interface will vary directly as the prevailing slip frequency at that location.

Further, conceive of each pocket working from north to south as having a slightly higher load-to-generation mismatch, meaning the north terminus is properly balanced but the south terminus is significantly undergenerated. Governors on all governed generators within each pocket are set for 4% speed droop, and setpoint of all governors is 60.00 Hz. "Power would flow North to South."

Frequency deviation from 60.00 Hz in Québec would be "Ontario at 59.87," due to proportional offset; similarly, "NY at 59.81, NJ at 59.76, Maryland at 59.65, North Carolina at 59.61, ect ect with the frequency dipping down to 59.00Hz at say Mississippi."

A workable scenario...only problem is, that's not the way it actually is in the real world; because the entire Eastern Interconnection is synchronously connected, there will not ever be any standing difference in frequency between any two interconnected parts of the interconnection.

Due to angular differences, though, there will for sure be real power transfers from North to South [not including Hydro-Québec / Trans-Énergie, which operates asynchronously, as previously explained]. Indeed there is almost always a transfer south at the Ontario-Michigan interface; there are four tie lines, all equipped with phase shifting transformers to allow for the balancing of flows.

Others have already touched on the implications of positive and negative ACE...[think Kosovo]

Hope this helps.

CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[waross]

Power would flow North to South and every generator would be turning slightly slower than the one above it. Not enough to motor or skip poles, but just enough to be lagging physically.

Hi Gunnar, Let's hit the pub. Like you, I'm off drink but I feel a desperate need for a tall Pepsi, Straight Up, in a Dirty Glass. I'm guessing that you could use something similar. First round's on me.

Historically a group of parallel generators would be operated in droop except for one swing set that corrected frequency errors for the system.
The base load sets used droop to control loading.
At a set point of 60 Hz, a generator would not be supplying any real energy to the system.
It may be overexcited to supply reactive power but that is a topic for another day.
To load a generator up, the set point would be advanced.
With 5% droop, if the frequency is set to 101% of 60 Hz, or 60.6 Hz the set will pick up 20% load.
The set is running at a nominal 60 Hz but the governor is set at 60.6 Hz.
At a governor setting of 102.5% of 60 Hz or 61.5 Hz the set will be running at a nominal 60 Hz, and 50% loaded.
For those with an instrumentation background think of 5% proportional band plus 5% offset. That's droop.
In the old plants there was a control marked UP-----Down or something similar. This was the governor control. When a generator was running but not paralleled this control controlled the speed/frequency. The operator could increase the speed/frequency with this control.
Once the generator was in parallel with the system, the control no longer controlled the speed/frequency. Those were locked in to the grid frequency.
Now the same control controlled the load on the set rather than the speed. What changed? Is there some fancy switch over circuit that transfers control from speed control to load control?
No. The control still controls the governor. Now as the governor is advanced, when the generator can not speed up, the droop or proportional action of the governor adds more fuel trying to speed up, it can't speed up and so takes on more load instead.
Now what happens when a large load hits the system? All the generators respond together and the system frequency drops a little due to the droop action of all the governors.
The controls on the swing set see that the frequency is no longer exactly 60 Hz and starts to pick up load to correct the frequency.
As the swing set is picking up the load, the frequency on the entire system is increasing until it stabilizes at 60 Hz or until there is another load change.

Another advantage to running in droop as well as the ease of loading sets is that in the event of loss of load for one or more generators, the droop control will limit the speed setting to 105%.
All the generators in a system run in synchronism at the same frequency.
Due to load changes, the frequency is continually varying a small amount and the swing set is continually correcting the errors in frequency.
We have seen some frequency graphs that show varying frequency on systems. It would be interesting to see the graph of the loading on a base set. The frequency chart and the load chart should mirror each other fairly well. (Yes, the scale will have to be adjusted to allow for a number of factors.)
That's the way it used to be.
I've been away from generation for a while now. Has there been any changes or is the same basic system used today?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[davidbeach]

Rather than thinking of those locations as different frequencies, think of them as different phase angles. Say there's a series of nodes from north to south that all have a 5 degree phase angle shift. Real power flow is north to south across every link. While it lasts that's "stable".

Now change something such that the new stable condition will be 10 degrees across every link (southern generation decreases or southern load increases, or some combination thereof). Some machines have to move 5 degrees, some have to move 10 degrees, some each of many different 5 degree increments. Just supposing, probably wouldn't be an even multiple of 5 in any real case. Now assume the whole system adjusts over some finite period of time. The units that need to move 5 degrees speed up almost not at all, but at each step the units have to all speed up a bit more. At the furthest reach there's a lot of speeding up to do, followed of course by settling down in to the new condition. During that time that they're all doing their own thing between this and that, that's when you'll find a variety of different frequencies on a single interconnected grid. Allow infinite time and it's easier to say everything stays at one frequency plus/minus some angular difference, but infinite time probably means instability.

There won't be standing differences in frequency, those are unsustainable. But at times there will be different frequencies, at least as determined by the relays measuring frequency at their point of connection to the system, and then those frequencies will all blend back together (ish). The problem is that while they're settling into this new state something else happens the results in the need for a new state.

The grid operator may measure frequency in a means that suggests small changes (at most) while the relays may measure frequency in a means that shows lots of variations across the whole interconnection, all different at different locations. Each equally valid in context.

Again, no standing differences in frequency, but lots of transitory differences in frequency. The whole 60.00, 59.87, 59.81, 59.76, 59.65, 59.61 thing won't happen over any appreciable amount of time. It just can't. Here I'm mostly in agreement with the "only one frequency" crowd.

As this thread began I was in the no standing differences camp, but when that became (or at least it seemed so to me) an exactly one frequency in any system at any time position I found myself on the side of multiple frequencies, but always as a transitory condition. I'm deliberately not using the word transient as it implies, at least to me, a much shorter time frame, cycles vs. seconds. The event happens, that's the transient. Following that transient the system is in a transitory state of flux, moving from one "stable" state to the next "stable" state. Lots of different things happening at different places, but all aimed at a new operating point. If nothing else happens the whole system would settle on a single new frequency, but something always happens.

I'm a relay guy, I'll go with frequency as measured by myriad relays through out the interconnection, many differences. Sometimes I wish the relays were more willing to accept a nominal frequency plus/minus, but no. I've done event reconstruction where I've had to account for minor variations in frequency between relays on the same bus in the same substation. It would be so nice to have ONE FREQUENCY, but that's not the real world.

I'm entirely open to somebody defining frequency in a logically consistent manner that washes away many of the differences that the relays see. I'm really surprised that nobody else has offered a definition, only insisted that frequency is frequency. Yeah, but...

If you're doing under frequency load shedding you need a specific definition of frequency. If you're doing time error correction (or actively ignoring that "antiquated concept" - see Kosovo above) then frequency means something else entirely different. ACE calculation probably needs an intermediate definition of frequency.

So, can we have a discussion of what a meaningful and useful definition of frequency might be, or does that need to be a separate thread? I think I'm rather more pragmatic than some of the response to my posts may have suggested.

What's your definition of frequency? How do you measure it? What do you use it for?

I'll offer two.

1) Frequency, at the plant level, is the electrical analog of instantaneous angular velocity of the (synchronous) machine shaft speed. RPM divided by the number of pole pairs divided by 60 (seconds per minute). Formulas that start with radians per second are equally valid.

2) Frequency, out on the system is the inverse of the time interval between successive positive zero crossings of a particular measured quantity, generally voltage, but could be current in a relay that doesn't have voltage inputs. A certain amount of data validation is required when using this method or weird things happen, see the NERC report on the Blue Cut Fire event. Relays I'm familiar with have a tendency to shut down frequency measurement following step changes that the real system can't make. They won't make the Blue Cut mistake but they can be the dickens to test for underfrequency tripping.

What's your definition and means of measurement?

 

[waross]

Thank you very much for taking the time and effort to share that information with us, David.
Does this work?
"Frequency, out on the system is the inverse of the time interval between successive positive zero crossings of a particular measured quantity, generally voltage, but could be current in a relay that doesn't have voltage inputs."
There shall be an exception for small apparent frequency errors across the grid caused by phase shift due to loading or unloading which shall be ignored.
If phase shifts due to loading are quantified and factored out will we then see one frequency across the grid?

I'm a relay guy, I'll go with frequency as measured by myriad relays through out the interconnection, many differences. Sometimes I wish the relays were more willing to accept a nominal frequency plus/minus, but no. I've done event reconstruction where I've had to account for minor variations in frequency between relays on the same bus in the same substation. It would be so nice to have ONE FREQUENCY, but that's not the real world.

I agree. But for that to happen, would the relays need a power flow input so as to be able to discriminate between real frequency changes and phase angle changes due to load changes?
Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[marks1080]

Definition of frequency?

As was previously mentioned, there are only two ways to measure this. I don't think that poster felt like they needed to specifically say you can measure either the voltage or the current waveform, but that is exactly what the poster was referring to. That is the definition of frequency. It's a fundamental concept, how else would you want to define this?

One of the confusing things I believe is that folks are assuming the primary system to always be sinusoidal. What myself and skogs and many others are poorly trying to say is that the waveform (voltage or current, pick either one you like) will be a single waveform for the entire primary system. If the system is nice a stable and being properly operated it'll be a nice sinusoidal very close to 60Hz with very little dynamic movement to the frequency. Regardless, it doesn't matter which part of the primary system you pick to measure, waveform period and frequency will always be identical because there is ONLY ONE WAVEFORM. This is ALWAYS true for a small system, or a large system.... the only difference being that its much easier to maintain the nice sinusoidal, 60hz, waveform for a large system. But the waveform, hence the frequency, will be identical no matter where you choose to measure on the primary part of the system.

It's nice to discuss what's going on at the low side of the generators and about phase angles, etc, and it does give people a better understanding of how GENERATION works.... HOWEVER, these things in no way could ever EVER EVER cause one part of the primary system to have a different voltage or current waveform than another part. The only part of the waveform that could change throughout the system is amplitude.

If you're reading minor variations in system frequency ON THE SAME BUS by different IEDs, it's a measurement limitation.

 

[wroggent]

In the Nov/Dec 2017 issue of IEEE Power & Energy there was an article about the rapidly growing amount of non-inertia generation on the grid. The article points out that as the amount of synchronous generation decreases by percentage, that inverter based generation will need to offer (be programmed for) frequency regulating functions that synchronous generators currently provide. The article also briefly mentions the challenge associated with this, in that inverter based generation measures the frequency from the grid whereas synchronous generation uses the shaft speed as a reference. Using zero crossings without any kind of filtering or intelligence can result in some pretty wildly varying measured 'frequencies', apparently.

 

[davidbeach]

The relays don't know, can't know, why any particular frequency change happens. A bunch of relays all connected to the same set of bus VTs will all see the same changes in frequency. A relay that's also seeing current on a heavily loaded line might someday have the smarts to recognize that the apparent change in frequency is due to shifting phase angles from changing load; but relays on lightly loaded lines, or seeing a constant load like a cap bank, won't know why the change. Even though they all have the same reference input, the relays all have measurement tolerances and may not all report precisely the same frequency.

If those phase shifts can be filtered out (averaging over multiple successive positive zero crossings for example) the regional differences get smaller and smaller. But if my UF load shedding relay has to make a trip decision in less than 11 cycles, it can't be looking a window much longer than a cycle. What makes sense as a frequency measurement for a local control center (ours is the average of multiple locations around the system) may be different than what makes sense for looking at a whole interconnection, and is certainly different from a relay level measurement.

The real exception to that zero crossing definition has to be the odd durations during faults. At the beginning of the fault you'll get one interval that bears no relationship to preceding or following intervals; likewise when the fault clears. Too many inverters impacted by the Blue Cut Fire events took a single long interval and decided that the frequency was really low and dropped out. Frequency hadn't changed that much, but the angular relationship between the phases certainly did. When the faulted line cleared the normal phase angles returned. A good frequency measuring device should just throw out those impossibly high or low results and wait for another cycle to try again. No step changes of frequency in systems with a decent amount inertia, just constant change.

 

[davidbeach]

One waveform is a great shorthand, and a useful approximation under many conditions, but is it always true, under every condition?

If there is always "ONLY ONE WAVEFORM" how do we explain waveforms close to faults that look very different from waveforms at the same instant far from the fault?

If there is always "ONLY ONE WAVEFORM" how do we explain what's going on during swings, including stable swings?

If there is always "ONLY ONE WAVEFORM" how do we explain inter-area oscillations or sub-synchronous resonances that occur in one part of the system and are not seen in other parts of the system.

 

[HamburgerHelper]

Add to that ERCOT's Reference System Operator Guide basically saying frequency change is required for changing power flows. Their guide recognizes frequency differences are necessary for changing power flows or changing voltage angles. The operators themselves are watching frequency and recognize that it is indicative of changing power flows.

 

[marks1080]

i give up. I think skogs already did.

Wait no i don't...

"If there is always "ONLY ONE WAVEFORM" how do we explain waveforms close to faults that look very different from waveforms at the same instant far from the fault?" Answer: Impedance

"If there is always "ONLY ONE WAVEFORM" how do we explain what's going on during swings, including stable swings?" Answer: This has nothing to do with comparing angles.

"If there is always "ONLY ONE WAVEFORM" how do we explain inter-area oscillations or sub-synchronous resonances that occur in one part of the system and are not seen in other parts of the system." Answer: again, we're not talking about angles between voltage and current waveforms. Just pick one and analyse it. Only one. Doesn't matter which one. Also: Impedance. - remember the topic here is frequency... and that's during stability. During instability it's more accurate to talk about the period, but even I wouldn't nit pick that much.

 

[davidbeach]

Then we’ll just have to agree to disagree. Over a second or two you are undoubtedly correct. Over a cycle or two I don’t think so. Relays don’t see phase angle shifts, they see frequency changes. Define your terms, you may have a definition that I can accept that let’s us both agree. But using the inverse of the time from one positive zero crossing to the next the relays will still measure different frequencies at different locations. The relays know nothing about phase shifts but they continuously calculate frequency.

Do you actually read what I write or do you just get upset with parts of it? I’ve never said there can’t be definitions of frequency that agree with your viewpoint. I’ve simply said that there are also valid definitions that contradict your viewpoint. Provide a definition of the measurement of frequency.

It’s gray. I’ll take a contrarian position to anyone who insists on either white or black. They’re both at odds with how the various devices that monitor and control the power system see the world. Like so many other things, the real answer is “it depends”. Insist it’s one and I’ll argue for the other. It’s neither and it’s both. View point matters. Definitions matter; I’ve offered a couple and hinted at others, what’s yours?

 

[marks1080]

*Edited out of respect for davidbeach*

Sorry the last reply was starting to veer off into unproductive territory. There's a tonne of valuable information in this thread, and it's been one of the better threads in a while so I really don't want to be the one who blows it up.

Apologies again.

 

[MatthewDB]

In the Nov/Dec 2017 issue of IEEE Power & Energy there was an article about the rapidly growing amount of non-inertia generation on the grid. The article points out that as the amount of synchronous generation decreases by percentage, that inverter based generation will need to offer (be programmed for) frequency regulating functions that synchronous generators currently provide.

There is already a bunch of work going on in this regards for small power systems. Islands that have diesel generators and PV have trouble for the same reason. Several solar inverter manufacturers have implemented algorithms already to do exactly that.

The article also briefly mentions the challenge associated with this, in that inverter based generation measures the frequency from the grid whereas synchronous generation uses the shaft speed as a reference. Using zero crossings without any kind of filtering or intelligence can result in some pretty wildly varying measured 'frequencies', apparently.

Only crappy ones that are going to have issues when there is even a minimal amount of distortion do something that simple. The typical method on an inverter is to run a phase locked loop (PLL) at some power of 2 multiple of the line frequency (often 128 or 256) and do an FFT of the line voltages. The PLL feedback is based on the phase error of the FFT voltages. Doing that filters out all but the line frequency distortion and gets highly accurate phase angle.

Present requirements are for inverters to quickly "go away" under disturbances because utilities are freaked out about islanding. That fear is unfounded though, as an inverter isn't a synchronous generator. It won't sit there off frequency and / or off voltage. As inverter penetration increases, the inverter manufacturers will help the grid once they get a green light from utilities.

 

[MatthewDB]

Dumb question - what is the impact of the speed of light? I would imagine that it acts like a "delay" to the disturbances traveling from one end of the system to the other.

 

[KHH1]

I'm new to the forum and also late to this discussion. It's an excellent topic with a lot of valuable information. Here is my understanding of the frequency in the AC power system:
- In synchronously connected AC system, there is one frequency. Every generator is tied to this frequency via the synchronizing torque
- If one measures the frequency at different locations, a slight difference in frequency measurements can be observed. Take a look at the real-time frequency of US in this link
Link

- Why there is slight variation: It is caused by local oscillation of the generator about the nominal speed. Since frequency is measured based on the voltage waveform, it is dependent on the local generators which are producing the waveform.

So what is the nature of this "local oscillation"? Some posters talk about subsynchronous oscillations, or electromechanical oscillation (local and inter-area modes). I believe it is caused by governor interaction. This type of interaction causes very low-frequency oscillations (~0.01 Hz) which can be observed across the system.

The frequency difference will become much more visible when there is a large disturbance.
Link

 

[HamburgerHelper]

KHH1,

Very cool. Thanks. There is more vatiation than I expected under normal operations.

The angle contour map is equally impressive. You can physically see all the wind in the midwest and I guess any other cheap generation supplying the northeast.

I am half tempted now to pull up a bunch of RPM data.

 

[wroggent]

Really neat khh1

 

[crshears]

Welcome to the fora, KHH1, and special thanks to you for providing Link 1. I would not have thought the frequency traces within the Eastern Interconnection would have routinely varied by as much as 0.01 Hz; I would have expected it to be of an order of magnitude less. As a consequence I'm beginning to think the participants in this group have indeed been describing the same elephant from opposite sides...

I do find it interesting though that when the Grid-Eye website map is showing the Eastern Interconnection in all red [ meaning 60.06 Hz ] the System Summary here at work is reporting our frequency as 60.02...maybe there is indeed some meter error out there...



CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]