Phase balancing and OHLE

[Irascible]

If you mean in terms of feeding power back into the grid, then the main issue is that grid frequency isn't fixed and instead varies slightly based on load.

I meant in terms of supplying the OHLE network rather than back into the grid - what I'd gathered from thread was the suggestion OHLE supply SFCs be synced to their inputs to present a completely in-phase output, but as you say that's a little variable. I did forget about regeneration, though ( although a) do they convert both directions?, and b) if they can at least convert to DC in reverse that's an obvious place for storage... ).

 

[najaB]

I meant in terms of supplying the OHLE network rather than back into the grid - what I'd gathered from thread was the suggestion OHLE supply SFCs be synced to their inputs to present a completely in-phase output, but as you say that's a little variable.

Two sections both fed by SFCs could be matched to each other using clocks (assuming no clock drift) but you'd still need a neutral section between them and any sections fed directly from the grid.

I did forget about regeneration, though ( although a) do they convert both directions?, and b) if they can at least convert to DC in reverse that's an obvious place for storage... ).

There's no reason that they can't be designed to convert in both directions since it just means converting the 25kV up to a higher AC voltage (which you could do with a 'dumb' transformer) and, given the importance of regenerative braking, I'd be very surprised if they aren't.

 

[edwin_m]

This is also one way that they can detect grid failure. They will try to set their frequency a little higher/lower than the grid input. If the grid is up they'll sense the current that leads to those 'bad things', if the grid is down then there won't be any opposing current. That's how pure grid-tie inverters know to shut down if the mains supply fails, and inverters with a battery backup will disconnect their input side to avoid back-feeding into the grid and potentially harming line workers.

Surely it's actually changing its phase? (which does involve changing the frequency but only for a very short time). If the frequency changes significantly, the inverter will soon be close to antiphase with the supply and those bad things can happen in a big way.

== Doublepost prevention - post automatically merged: 25 Jun 2024 ==

Two sections both fed by SFCs could be matched to each other using clocks (assuming no clock drift) but you'd still need a neutral section between them and any sections fed directly from the grid.

There's no reason that they can't be designed to convert in both directions since it just means converting the 25kV up to a higher AC voltage (which you could do with a 'dumb' transformer) and, given the importance of regenerative braking, I'd be very surprised if they aren't.

Yes clocks could be used but there would need to be some way of keeping them synchronised. I guess GPS time would do this sufficiently accurately, though it would stop working in the event of GPS spoofing/jamming or Kessler syndrome.

I don't believe "traditional" 25kV feeder stations regenerate back into the grid - they feed a large area and with relatively little voltage drop there's usually something somewhere that can use the power whenever a train regenerates. If there isn't, it is dumped into on-train resistors. An SFC probably feeds a smaller area (though this would be larger if it was synchronised to others without intervening neutral sections) so perhaps it's more worthwhile to make them regenerate back into the supply. In that configuration they are very similar to a modern traction package, converting between single phase and three phase via a DC link.

 

[najaB]

Surely it's actually changing its phase? (which does involve changing the frequency but only for a very short time). If the frequency changes significantly, the inverter will soon be close to antiphase with the supply and those bad things can happen in a big way.

My experience is with Victron inverters so others may do it differently, but this is the way that they explain it:

To detect a loss of mains, the inverter/charger will constantly try to shift the AC frequency. When connected to a stable grid, with a normal accepted impedance, it will not be possible to do so, and it can therefore detect that the mains is still present. In case the mains was lost, the inverter/charger will be able to shift the frequency at will, without any resulting correction currents.

Link: https://www.victronenergy.com/live/ve.bus:grid-codes-and-loss-of-mains-detection

I suppose the distinction between phase shift and frequency is effect vs cause.

 

[edwin_m]

My experience is with Victron inverters so others may do it differently, but this is the way that they explain it:

Link: https://www.victronenergy.com/live/ve.bus:grid-codes-and-loss-of-mains-detection

I suppose the distinction between phase shift and frequency is effect vs cause.

The operative words in your quote are "try to". As soon as the frequency shifts a tiny bit the inverter will start to drift out of phase, creating what is in effect a partial short circuit.

 

[najaB]

The operative words in your quote are "try to". As soon as the frequency shifts a tiny bit the inverter will start to drift out of phase, creating what is in effect a partial short circuit.

And that's what I said in my original post:

This is also one way that they can detect grid failure. They will try to set their frequency a little higher/lower than the grid input.

So I think we are saying the same thing, no?

 

[Tim_UK]

As I'm not a power engineer, is there anything impractical about syncing SFC output using a clock rather than their input?

The mobile phone networks do this by receiving clock synchronisation. Traditionally over the backhaul network, but more recently they use gps and other gnss networks. Look for the tiny antennas on top of base stations.

Mobile networks are looking to sync within 3 micro seconds. That seems like plenty to sync 2 AC sources.

 

[Irascible]

The mobile phone networks do this by receiving clock synchronisation. Traditionally over the backhaul network, but more recently they use gps and other gnss networks. Look for the tiny antennas on top of base stations.

Mobile networks are looking to sync within 3 micro seconds. That seems like plenty to sync 2 AC sources.

GPS/GNSS in general also rely on highly accurate clocks ( or it just doesn't work - even has to deal with relativistic effects ), I'd imagine GSM-R carries a timing signal, the source wouldn't be a problem - I was more curious about things related to managing the output & whether you'd want them all synced precisely or you need one to ( eg ) sync to it's neighbour's phase because of imprecise distances between them, or something along those lines.

 

[edwin_m]

And that's what I said in my original post:

So I think we are saying the same thing, no?

I think we are setting out the same thing, but I suggest the way describe it is a little confusing. Engineers would describe the action as a phase change, even though it needs a small change in frequency (then return to the original frequency very soon after) to achieve it. In general there seems to be a bit of mis-understanding on this thread - inverters feeding a shared circuit such as an OLE section can't be running at different frequencies, and can't be out of phase except by a very small amount.

The mobile phone networks do this by receiving clock synchronisation. Traditionally over the backhaul network, but more recently they use gps and other gnss networks. Look for the tiny antennas on top of base stations.

Mobile networks are looking to sync within 3 micro seconds. That seems like plenty to sync 2 AC sources.

That probably means that if Putin messes with GPS, or satellites are brought down by a cascade of space debris (Kessler syndrome) our mobiles will all go down too.

 

[AM9]

I think we are setting out the same thing, but I suggest the way describe it is a little confusing. Engineers would describe the action as a phase change, even though it needs a small change in frequency (then return to the original frequency very soon after) to achieve it. In general there seems to be a bit of mis-understanding on this thread - inverters feeding a shared circuit such as an OLE section can't be running at different frequencies, and can't be out of phase except by a very small amount.

That probably means that if Putin messes with GPS, or satellites are brought down by a cascade of space debris (Kessler syndrome) our mobiles will all go down too.

Not necessarily, - there is probably a mode where asynchronous or base to base sychronisation of the system can be established under such conditions, simuilar to reversionary mode that avionics systems have.

 

[Irascible]

Or you can just do it all by landline if you have to.

 

[najaB]

Or you can just do it all by landline if you have to.

Laughing at the mental image of two electrical control operators trying to synchronise their sections by calling the speaking clock.

 

[AM9]

Or you can just do it all by landline if you have to.

I wasn't suggesting radio but any reliable comms link could carry a local sync signal. In a breakdown situation though, a radio link is probably more likely.

 

[Irascible]

Laughing at the mental image of two electrical control operators trying to synchronise their sections by calling the speaking clock.

Don't even start, you can bet someone somewhere will suggest AI calling it...

BT's fibre backbone might be a good candidate for clock feeds to anywhere that needs it robustly. Some sort of is likely the first choice though, yes.

 

[poffle]

Don't even start, you can bet someone somewhere will suggest AI calling it...

BT's fibre backbone might be a good candidate for clock feeds to anywhere that needs it robustly. Some sort of is likely the first choice though, yes.

Historically synchronisation in the telecoms networks has been a big issue ever since digital transmission/switching was introduced in the 1980s. The top-level exchanges had Caesium clocks ( these are really standard frequency sources rather than hour/minutes/seconds clocks ). These produced a 2MHz output which was then distributed around the network on 2Mb/s links which were the standard building blocks of the network.

On a PSTN call in the 1980s you could get regular clicks on a call when two analogue exchanges were connected over a digital transmission system where the system was incorrectly configured and both ends were operating in their own built in crystal clock.

Nowadays I think mobile networks tend to use GPS receivers a lot for synchronisation.

 

[172007]

A grid-tied inverter MUST be locked for frequency and phase when exporting, just as a rotating alternator must. Otherwise bad things will happen. It is a trivial matter to make the inverter's 50 Hz oscillator match the external utility supply in frequency and phase.

I am no expert but apparently it's an issue with small groups of houses or villages that are an island as they are on a spur after being stepped down from higher voltage. Apparently if too many properties have solar the frequency can gradually creep up.

 

[MadMac]

The Metro Blue Line in the Los Angeles area has synchronized 100 Hz supplies for the AC track circuits which also provide the speed codes to the trains - it’s a very similar arrangement to the early LU Victoria Line setup. It was, shall we say, “entertaining“ if one of the “master” supplies failed!

 

[QueensCurve]

There was/is a technique invovling a "T transformer", which lets you turn three phases 120 degrees out of phase into two phases 90 degress out of phase while conserving balance so you only have to split your system into two peices rather than three. Not sure if this technique is used in the UK..

I hadn't heard ot the T transformer before. There is a description at https://www.electricaltechnology.org/2022/03/scott-t-connection-transformer.html

They're much more common in Germany where the railway supply is 16.7Hz and you have to convert the frequency. In the past this was done with motor-generator sets, where a 50Hz three-phase motor with three windings around the stator drove a shaft at 16.7rpm, which in turn drove a single-phase generator at 16.7Hz. It's all done with solid-state power electronics now, of course.

The 15kV low frequency system was pioneered in Switzerland by the Bern Lötschberg Simplon railway which opened in 1913 and took its supply from rotary frequency converters of the Forces Motrices Bernoises (FMB).

 

[poffle]

I am no expert but apparently it's an issue with small groups of houses or villages that are an island as they are on a spur after being stepped down from higher voltage. Apparently if too many properties have solar the frequency can gradually creep up.

That would be voltage creeping up rather than frequency. If the voltage exceeds a threshold then the inverters on the solar panels would stop feeding into the distribution network.

 

[najaB]

That would be voltage creeping up rather than frequency. If the voltage exceeds a threshold then the inverters on the solar panels would stop feeding into the distribution network.

I was trying to think of a reason that the frequency would increase, voltage rising makes more sense.

The reason this happens is because the solar inverters generate at a slightly higher voltage than the grid supply so that the current can flow "up hill" so to speak back into the grid. The issue is that if there's more power being generated than used on a distribution circuit then the inverters end up chasing a higher and higher voltage as they all fight to export power.

 

[stuving]

There is a very thorough treatment of the issues around phase balance in: Engineering Recommendation P24, Issue 2 2020, AC supplies to railway systems. However, be warned - it is written primarily for power transmission experts, though with explanations to assist us mere mortals. This revision is from 2020, so while SFCs are mentioned they are not covered in the recommendations - the text notes that none were in use then.

You do have to watch out for things that look clear enough but are in fact used in a jargon sense that you just have to know. For example, feeds are described as 1x25kV and 2x25kV, which looks like one circuit (and transformer) or two. But it does not - it refers to two-wire (for BT systems) and three-wire (25-0-25kV for AT systems) circuits.

The Scott transformer is covered, but as an alternative, not a recommended practice. Its drawback is that it only gives a balanced load on the three-phase side if the two single-phase loads are equal. But traction loads are quite variable, so two adjacent sections will draw significantly different powers quite often. Account must also be taken of outages, planned and unplanned, which may remove one section's load altogether. I suspect there may also be a loss of switching flexibility for alternative feed arrangements to work round outages.