Stability of grids with a high percentage of asynchronous sourcing

[Phil1977]

But this is what I've been claiming all along: you will need devices to govern the grid stability regardless how they are physically constructed. Decentralised generation or not, there have to be devices which govern synchronisity. Like an orchestra needs a director. But those grid synchronising devices will need to be loud enough to overcome the noise as each micro inverter feeding the grid adds noise due to component tolerances and measuring errors.

And for sure you don't need an atomic clock perse, but it certainly is a nice to have as even today clocks depend on the grid frequency being stable most of the time.

The comparison with an orchestra and a director is nice - because (as far as I know) the instruments in an orchestra more or less synchronize themselves too. They don't look at the director to know which note has to be played. But the director gives some advice and guidance to make everything perfect.

In the grid that means: You don't need very big central instances for the buffering. Thousands of medium sized inverters can do the trick together. It doesn't matter if you have one large mass of inertia or thousand small ones.

Of course you need some supervising power - each grid has some central office that plays the director and does the checks and balances. For this purpose it´s definitely good to have precise measuring instruments incl. precise clocks. And no one seriously doubts supervising a grid is complicated - I think not even whole EEVBlog will know everything you have to care about there.

But luckily it´s just not true that our wealth and grids purely rely on coal or uranium hungry energy monsters.

[Andy Chee]

The comparison with an orchestra and a director is nice - because (as far as I know) the instruments in an orchestra more or less synchronize themselves too. They don't look at the director to know which note has to be played. But the director gives some advice and guidance to make everything perfect.

And incidentally, any one who has experience of conducting an orchestra or choir on the first day of rehearsals, would know how unstable they can be!

[Someone]

This device has a large rotating mass that acts as a kind of flywheel with a certain level of rotational energy. If voltage or current in the net are out of phase, or if load spikes need to be dampened then energy is fed into or taken from this flywheel.

It´s really not difficult to setup a good physical model when electricity needs to be converted into rotation or when it´s necessary to generate electricity from rotation. You can take exactly these rules to build a circuit that stores or takes energy from a big capacitor.

This is - extremely simplified - what's done in grid stabilizing inverters. You have a well buffered DC bus that can feed the grid with power exactly when a beautiful sine wave needs it - or even take momentary power in certain circumstances.

But this is what I've been claiming all along: you will need devices to govern the grid stability regardless how they are physically constructed. Decentralised generation or not, there have to be devices which govern synchronisity. Like an orchestra needs a director. But those grid synchronising devices will need to be loud enough to overcome the noise as each micro inverter feeding the grid adds noise due to component tolerances and measuring errors.

And for sure you don't need an atomic clock perse, but it certainly is a nice to have as even today clocks depend on the grid frequency being stable most of the time.

A rather long bow to draw, just to inject your disingenuous points.

Noise is not a problem when it is adding together a large number of unbiased noise sources, that averages out to "nothing". If any specific generator has a local clock frequency error, all that adds is a small change in the power it is delivering to the grid. Which is fairly metered, and the only problem is that generator might be contributing (and being paid for) less energy than they otherwise would be entitled to generate. No grid stability (that inertia could protect from) is lost from inaccurate clocks, only long term frequency accuracy could be affected and thats assuming there was zero action taken by the grid operators (which is unbelievable).

Using a non-specific "stable" without any qualification of what time scale that is continues your patterns of dropping half an argument and coming back to point out how people mis-interpreted you. Troll behaviour. Grid instability (inertia relevant timescales of single cycle to seconds) from inaccurate clocks would require something unbelievable like sudden/instantaneous multiple percent changes appearing in local oscillators, implausible and if that was normal/typical for the device it would fail compliance testing before being deployed.

Synchronicity (an antonym to what you likely mean): "the simultaneous occurrence of events which appear significantly related but have no discernible causal connection". Rather the grid synchronisation is by design and a simple causal relationship.

[CatalinaWOW]

Haven't read every line of this whole thread, but it seems to me there are some missed points and clarification needed.  Referring to a concept from early in the thread, I am not aware of any fundamental issues that prevent a solution.  But there are very real implementation issues, most of which have theoretical solutions not all of which are fully implemented as renewables exist today.

Issue 1.  As the solar system in my home is configured (and probably most grid tied systems in the world), when there is a fault on the incoming line the system disconnects from the grid.  This is done for the safety of grid workers as they repair the fault.  It is bad for linemen when they find a "disconnected" line energized by attached renewable sources.  This doesn't cause any really serious problems when renewables constitute a small fraction of total generation.  But when that proportion becomes significant there are various issues.  When the fault is due to one of the remaining large singular generation sources failing the effect is magnified by the loss of all the attached renewable.  And there are startup problems.  There are lots of ways to solve this problem, but to my knowledge there isn't much effort yet on implementation.  As mentioned previously, no one wants to pay for it.

Issue 2.  DC grids are not automatically stable.  The issues are different, again all solvable AFAIK.  But there has been next to no investment in DC grids with more than a few dozen sources and sinks.  Even fewer where operations are split over a huge number of owners and organizations.

Issue 3.  Something I am surprised that no one in this forum has brought up is filtering for the noise caused by power lines.  I have several pieces of equipment that have notch filters for power line frequency.  If power line frequency is allowed to drift widely from current norms these filters will be noticeably less effective.  There are other effects from allowing large drifts in frequency, mostly involving changes in line loss, reactive power and core loss in equipment that is often carefully optimized for one frequency.  Equipment from a 50 Hz network usually works on 60 Hz and visa versa, but not as well.  Clocks are not the only reason to keep power line frequency relatively constant. 

[Andy Chee]

Speaking of synchronisation:



The metronomes are analogous to conventional spinning turbines due to their pendulum weight providing the equivalent of inertia.

The metronome equivalent of an inverter would be a HDD voice coil head ticking back & forth. i.e. no inertial mass, and would rely on sensors to detect and trigger the voice coil.

[Someone]

Haven't read every line of this whole thread

Well perhaps you should?

Issue 3.  Something I am surprised that no one in this forum has brought up is filtering for the noise caused by power lines.  I have several pieces of equipment that have notch filters for power line frequency.  If power line frequency is allowed to drift widely from current norms these filters will be noticeably less effective.  There are other effects from allowing large drifts in frequency, mostly involving changes in line loss, reactive power and core loss in equipment that is often carefully optimized for one frequency.  Equipment from a 50 Hz network usually works on 60 Hz and visa versa, but not as well.  Clocks are not the only reason to keep power line frequency relatively constant.

No-one is seriously suggesting frequency deviations larger than already exist:
https://www.eevblog.com/forum/chat/will-ac-power-distribution-become-obsolete/msg5550023/#msg5550023
The existing grid will basically collapse (by design) when frequency is off by a few percent, that is considered gross fault. Load shedding (emergency intervention) is usually starting at 1% deviation.

Issue 1.  As the solar system in my home is configured (and probably most grid tied systems in the world), when there is a fault on the incoming line the system disconnects from the grid.  This is done for the safety of grid workers as they repair the fault.  It is bad for linemen when they find a "disconnected" line energized by attached renewable sources.  This doesn't cause any really serious problems when renewables constitute a small fraction of total generation.  But when that proportion becomes significant there are various issues.  When the fault is due to one of the remaining large singular generation sources failing the effect is magnified by the loss of all the attached renewable.  And there are startup problems.  There are lots of ways to solve this problem, but to my knowledge there isn't much effort yet on implementation.  As mentioned previously, no one wants to pay for it.

Anti islanding is (almost entirely) administrative/regulatory control demanded by the grid operators. Grid operators have and do license inverter systems to provide virtual inertia but so far this has only been larger "centralised" plants which can be commanded like existing generators. 99% of the problem preventing inverters participating in inertia is grid operators demanding extremely fast anti-islanding.

Non-trivial grids have already operated with insignificant amounts of mechanical inertia:
https://spectrum.ieee.org/electric-inverter
That an inverter can be configured to emulate the dynamics of existing plant (which they can) is all that is required for them to incrementally replace existing mechanic inertia. Known/proven/deployed already, there is no question of "can" inverters provide grid stability, it is only a question of how they will be integrated/compensated/economical.

[zilp]

The metronomes are analogous to conventional spinning turbines due to their pendulum weight providing the equivalent of inertia.

The metronome equivalent of an inverter would be a HDD voice coil head ticking back & forth. i.e. no inertial mass, and would rely on sensors to detect and trigger the voice coil.

I mean, that's kinda a useful analogy ... but also not. Because the important question is not whether there are "sensors" involved. The important question is whether the bandwidth of the regulation loop is sufficient. And there is no fundamental reason why things should be too slow just because electricity is involved.

After all, if you were to use a voice coil with some high resolution position encoder and high-bandwidth error amplifier, you can get it to be pretty stiff, too. AIUI, that's not what modern HDDs do, probably because they need extremely high positioning accuracy ... but then, you don't need to control mains voltage to microvolts.

So, it's not necessarily a bad analogy, but also, it doesn't really give any useful insight into the stability of inverter driven grids.

[Phil1977]

The comparison with an orchestra and a director is nice - because (as far as I know) the instruments in an orchestra more or less synchronize themselves too. They don't look at the director to know which note has to be played. But the director gives some advice and guidance to make everything perfect.

And incidentally, any one who has experience of conducting an orchestra or choir on the first day of rehearsals, would know how unstable they can be!

And no one says that first studies with grid forming inverters did not release lots of magic smoke and have blown a lot of fuses.

But these rehearsals proof now successful for quite a long time.

As far as I know there was no black start of a really large grid like North America or Europe yet that was initiated by inverters. But that´s probably not necessary at all in future. Here in Germany we usually plan with hydroelectricity and gas turbines to start the segments of the net and then incrementally switch on loads and other generators. These plants won't vanish from large grids very soon.

How to re-synchronize different areas of the grid after loss of synchronization is another complicated topic, it´s practised not often but it was always done reliably.

[Phil1977]

Issue 2.  DC grids are not automatically stable.  The issues are different, again all solvable AFAIK.  But there has been next to no investment in DC grids with more than a few dozen sources and sinks.  Even fewer where operations are split over a huge number of owners and organizations.

That´s a really interesting topic. DC grids for long distance transfer seem very tempting regarding efficiency and possible power levels. But investments to get this stable would be huge, no one wants the be the first. That makes it even less desirable for intermediate distribution levels.

Transformer based AC grids are there, are well-known and physically quite rigid. It´s very understandable no one wants to abandon such a well working system. But maybe in a distant future with certain countries exporting huge amounts of PV or wind energy, maybe then a big high voltage DC bus gets attractive to build.

[CatalinaWOW]

Haven't read every line of this whole thread

Well perhaps you should?

Issue 3.  Something I am surprised that no one in this forum has brought up is filtering for the noise caused by power lines.  I have several pieces of equipment that have notch filters for power line frequency.  If power line frequency is allowed to drift widely from current norms these filters will be noticeably less effective.  There are other effects from allowing large drifts in frequency, mostly involving changes in line loss, reactive power and core loss in equipment that is often carefully optimized for one frequency.  Equipment from a 50 Hz network usually works on 60 Hz and visa versa, but not as well.  Clocks are not the only reason to keep power line frequency relatively constant.

No-one is seriously suggesting frequency deviations larger than already exist:
https://www.eevblog.com/forum/chat/will-ac-power-distribution-become-obsolete/msg5550023/#msg5550023
The existing grid will basically collapse (by design) when frequency is off by a few percent, that is considered gross fault. Load shedding (emergency intervention) is usually starting at 1% deviation.

Issue 1.  As the solar system in my home is configured (and probably most grid tied systems in the world), when there is a fault on the incoming line the system disconnects from the grid.  This is done for the safety of grid workers as they repair the fault.  It is bad for linemen when they find a "disconnected" line energized by attached renewable sources.  This doesn't cause any really serious problems when renewables constitute a small fraction of total generation.  But when that proportion becomes significant there are various issues.  When the fault is due to one of the remaining large singular generation sources failing the effect is magnified by the loss of all the attached renewable.  And there are startup problems.  There are lots of ways to solve this problem, but to my knowledge there isn't much effort yet on implementation.  As mentioned previously, no one wants to pay for it.

Anti islanding is (almost entirely) administrative/regulatory control demanded by the grid operators. Grid operators have and do license inverter systems to provide virtual inertia but so far this has only been larger "centralised" plants which can be commanded like existing generators. 99% of the problem preventing inverters participating in inertia is grid operators demanding extremely fast anti-islanding.

Non-trivial grids have already operated with insignificant amounts of mechanical inertia:
https://spectrum.ieee.org/electric-inverter
That an inverter can be configured to emulate the dynamics of existing plant (which they can) is all that is required for them to incrementally replace existing mechanic inertia. Known/proven/deployed already, there is no question of "can" inverters provide grid stability, it is only a question of how they will be integrated/compensated/economical.

Perhaps you know that the person who suggested that deviations of ten Hertz in this thread didn't matter wasn't serious.  I don't.  And perhaps this comment will tell you that I did read almost every line.

I will repeat that I agree that all of the problems have solutions.  But your comment that the problems with islanding are legislative and therefore some not real is strange.  A non-technical problem is still a real problem, and will require time and investment to correct.  Investment in changing the laws and investment in retrofits/adjustments to the the growing existing infrastructure.

[coppice]

But your comment that the problems with islanding are legislative and therefore some not real is strange.  A non-technical problem is still a real problem, and will require time and investment to correct.  Investment in changing the laws and investment in retrofits/adjustments to the the growing existing infrastructure.

But islanding is a technical problem. If an island forms you can't shut down the power for maintenance work, so you have a massive safety problem.... or is fried technicians not a technical problem? They could wear extra PPE, I guess. The legislation specifically responded to a technical problem.

[Phil1977]

This islanding is an organizational problem in the worst case.

The small inverters you have in a single buildings usually don't generate much more power than they consume. They do not need to care for grid stability.

What it´s all about are the medium sized generating units. Like "agricultural" solar farms, wind farms, larger combined heat and power - just by gut feeling I´d say everything >500kW. These units need to be managed in a way that they form a stable power backbone.

Probably you can limit this inverter quality to everything connected to the intermediate voltage level grid. And there you need strict organizational safety measures to ensure safe conditions for any kind of technician in any case.

[coppice]

This islanding is an organizational problem in the worst case.

How is the power not being guaranteed to stop when the grid operator turns it off for maintenance or an emergency an organisational problem? Is your idea of organisational that maintenance only occurs at night, when the sun doesn't shine, and every source can be expected to drop out? Battery storage has made even that a bit dodgy.

[CatalinaWOW]

Islanding is a technical problem.  The legislative problem is that the solution it forced is driven only by safety, the legislation does not prepare for large amounts of small solar systems.

I believe my system is somewhat typical.  It is designed to draw zero energy from the grid when averaged over a one year interval.  Which means there is a large excess of power generated during peak months.  And while the system is trivial in size from the standpoint of a utility (peak output is around 12kW) penetration of these small systems is growing rapidly.  I am sure there are hundreds within 50 kilometers of me, and if penetration grows to the 20-50% level you will be talking tens of thousands of these small systems with total peak capacity measured in megawatts.

Whether this kind of growth actually occurs is unclear.  With current pricing (both on the solar side and utility rate side) it makes economic sense to install solar, but the high upfront costs and relatively long payback interval definitely is a barrier to adoption. 

But the real barrier is the same one driving utility scale adoption of solar.  If renewables are ever to be a huge part of the total electrical system the seasonal availability of these sources needs to be addressed.  There are three solutions I am aware of;  sizing the system (solar or wind) to meet requirements in the worst case season, developing season level storage capacity, or developing transcontinental grids some part of the grid is always in favorable conditions.  The investment for none of those approaches is currently planned.  At my location the available solar in the dead of winter drops to 10% or less of the peak summer power.  That level of increased capacity makes solar dramatically more expensive than even nuclear.  Seasonal storage makes that option look cheap.  Transcontinental grids appear to be the cheapest solution, but I see no planning for this approach, and little prospect for the level of international cooperation it would require.

[tszaboo]

There is no central reference voltage or frequency on the grid. It is distributed amongst all the generators. In the UK typical consumption now is about 30GW, so even the largest 1GW generator is very small, so no individual generator can vary either voltage or frequency. This is what is called an infinite busbar.

Never mind the central frequency issue.
How do you restart the grid in case of a backout? Imagine a cable was cut, and parts of the country goes black.
Instead of 20 power plants, that will generate electricity on command, you have a million inverters. All of them will say that they are in islanding mode, and don't generate electricity. You need to increase the voltage on the line (while load is still connected) to 200V or so for the inverters to start exporting electricity.
How do you do that? Maybe we should keep a few nuclear power plants just in case.

[nctnico]

There is no central reference voltage or frequency on the grid. It is distributed amongst all the generators. In the UK typical consumption now is about 30GW, so even the largest 1GW generator is very small, so no individual generator can vary either voltage or frequency. This is what is called an infinite busbar.

Never mind the central frequency issue.
How do you restart the grid in case of a backout? Imagine a cable was cut, and parts of the country goes black.
Instead of 20 power plants, that will generate electricity on command, you have a million inverters. All of them will say that they are in islanding mode, and don't generate electricity. You need to increase the voltage on the line (while load is still connected) to 200V or so for the inverters to start exporting electricity.
How do you do that? Maybe we should keep a few nuclear power plants just in case.

Well, the scenario for this would be to start with one primary, grid forming source (fossil, wind, solar, hydro, nuclear or storage; doesn't matter for as long as it is big enough) and bring up the other primary (grid forming) sources one by one so they can all align their phase & frequency. Once the core grid is powered up and stable, then connect the consumers gradually. At that point the grid tracking sources can start pumping energy into the grid.

[zilp]

But the real barrier is the same one driving utility scale adoption of solar.  If renewables are ever to be a huge part of the total electrical system the seasonal availability of these sources needs to be addressed.  There are three solutions I am aware of;  sizing the system (solar or wind) to meet requirements in the worst case season, developing season level storage capacity, or developing transcontinental grids some part of the grid is always in favorable conditions.  The investment for none of those approaches is currently planned.  At my location the available solar in the dead of winter drops to 10% or less of the peak summer power.  That level of increased capacity makes solar dramatically more expensive than even nuclear.  Seasonal storage makes that option look cheap.  Transcontinental grids appear to be the cheapest solution, but I see no planning for this approach, and little prospect for the level of international cooperation it would require.

Over here, we have more renewable enery in the winter than in the summer, because wind is generating so much more in the winter.

So, no, you don't need "seasonal storage", at least not in the sense of "storing all the energy you need for winter", and you also don't need to overprovision solar by a factor of 10. You just need a good mix of sources plus some storage.

[coppice]

But the real barrier is the same one driving utility scale adoption of solar.  If renewables are ever to be a huge part of the total electrical system the seasonal availability of these sources needs to be addressed.  There are three solutions I am aware of;  sizing the system (solar or wind) to meet requirements in the worst case season, developing season level storage capacity, or developing transcontinental grids some part of the grid is always in favorable conditions.  The investment for none of those approaches is currently planned.  At my location the available solar in the dead of winter drops to 10% or less of the peak summer power.  That level of increased capacity makes solar dramatically more expensive than even nuclear.  Seasonal storage makes that option look cheap.  Transcontinental grids appear to be the cheapest solution, but I see no planning for this approach, and little prospect for the level of international cooperation it would require.

Over here, we have more renewable enery in the winter than in the summer, because wind is generating so much more in the winter.

So, no, you don't need "seasonal storage", at least not in the sense of "storing all the energy you need for winter", and you also don't need to overprovision solar by a factor of 10. You just need a good mix of sources plus some storage.

What an incredible dumb thing to say. First you refer to your own circumstances, then you try to generalise what works for you to be something that works for others. RENEWABLES DON'T GENERALISE. Every spot on the planet brings its own benefits and challenges.

[zilp]

What an incredible dumb thing to say. First you refer to your own circumstances, then you try to generalise what works for you to be something that works for others. RENEWABLES DON'T GENERALISE. Every spot on the planet brings its own benefits and challenges.

It's just that this generalizes relatively well. Yeah, of course you don't have more wind during winter at the equator ... because you don't have winter.

But also, I was specifically responding to CatalinaWOW, who happens to have a US flag next to the name, and for large parts of the US, this should apply.

Also, the point wasn't primarily that everyone has more renewables during winter, but that wind tends to be negatively corralated with sunshine, so it's not like you can't have renewable power when the sun isn't shining ... to the point that at least in some places, you actually have more renewables during the winter, despite what many people believe.

Really, hydro power is way more variable across the globe.

[coppice]

What an incredible dumb thing to say. First you refer to your own circumstances, then you try to generalise what works for you to be something that works for others. RENEWABLES DON'T GENERALISE. Every spot on the planet brings its own benefits and challenges.

It's just that this generalizes relatively well. Yeah, of course you don't have more wind during winter at the equator ... because you don't have winter.

But also, I was specifically responding to CatalinaWOW, who happens to have a US flag next to the name, and for large parts of the US, this should apply.

Also, the point wasn't primarily that everyone has more renewables during winter, but that wind tends to be negatively corralated with sunshine, so it's not like you can't have renewable power when the sun isn't shining ... to the point that at least in some places, you actually have more renewables during the winter, despite what many people believe.

Really, hydro power is way more variable across the globe.

Having the US flag beside his name means he could live in absolutely any kind of climate. Desert. Perpetual ice. No wind. Endless wind. No rain. Endless rain. You name it, and some part of the US experiences it.