Electric power systems: secondary networks

[AlbertL]

There have been several threads discussing various aspects of power distribution, but one interesting topic I haven't seen is secondary networks.

The most common distribution system, called a "radial" system, consists of a feeder circuit (generally in the 4-35 kV range) originating at a substation and supplying multiple transformers, each of which serves one or more customers.  The key characteristic of this system is that any given customer is supplied by only one feeder circuit.   

In a secondary network, customers are served by two or more transformers paralleled on the secondary side and supplied by different feeder circuits, possibly from different substations.  The simplest form is the "spot" network, having only one customer on the secondary bus.  In an "area" network, the bus serves multiple customers.

A key component of a secondary network is the network protector.  This is a form of circuit breaker inserted between the secondary of each transformer and the bus.  Its purpose is to protect the system in case of a short circuit ("fault") or outage on a feeder by preventing reverse power flow from the secondary bus through the transformer into the primary feeder.     

[richard.cs]

The most common distribution system, called a "radial" system, consists of a feeder circuit (generally in the 4-35 kV range) originating at a substation and supplying multiple transformers, each of which serves one or more customers.  The key characteristic of this system is that any given customer is supplied by only one feeder circuit.

In the UK we have distribution circuits like this at 11 and 33 kV (phase-phase), but ours tend to be a bit different. In most cases the feeder forms a ring from the primary substation, around some number (perhaps 10-15) secondary substations each with 1 or 2 transformers supplying a few hundred customers each, and back to the primary substation. The ring is broken at one point, and is fed from both ends essentially creating two radial circuits. The nice feature though is that the break can be remotely switched to occur at any of the secondary substations, or at two of them, such that any one faulty length of cable can be isolated for repair without loosing the 11 kV feed to any of the secondary substations.

In a secondary network, customers are served by two or more transformers paralleled on the secondary side and supplied by different feeder circuits, possibly from different substations.  The simplest form is the "spot" network, having only one customer on the secondary bus.  In an "area" network, the bus serves multiple customers.

We do have substations with multiple transformers and a common busbar, but so far as I am aware it's more common to have the busbar split into sections rather than operate transformers in parallel in normal operation. We commonly have LV feeders where each end connects to two secondary substations, which may or may not be on the same 11 kV feed. Most often these are connected as radial feeders, fed from one end only or fed from both with a break in the middle, but in areas with very high load density (central London for instance) the network is sometimes operated meshed with two or more substations feeding in. This gives rise to very high levels of fault current, and exciting failure modes if HV is lost at one site but not another. It gets interesting when a sizeable area ends up with its 11 kV supply powered via a distribution transformer operating in reverse and a bit of very hot-running LV cable down the street from the next substation.

A key component of a secondary network is the network protector.  This is a form of circuit breaker inserted between the secondary of each transformer and the bus.  Its purpose is to protect the system in case of a short circuit ("fault") or outage on a feeder by preventing reverse power flow from the secondary bus through the transformer into the primary feeder.     

I don't have a huge amount of experience with UK meshed networks (only a few towns have them), but as I understand it in most cases this protection is not provided and reverse power flow is allowed to persist until remote network reconfiguration can be used to restore power to the faulted section or until simple overload protection operates. Withstanding short-term operation like this is part of the design.

[NiHaoMike]

A key component of a secondary network is the network protector.  This is a form of circuit breaker inserted between the secondary of each transformer and the bus.  Its purpose is to protect the system in case of a short circuit ("fault") or outage on a feeder by preventing reverse power flow from the secondary bus through the transformer into the primary feeder.     

How does that work when there's solar or other grid tie energy sources on the customer side? Is the reverse power breaker set to trip only if the backflow is greater than the sum of all distributed energy generation in that zone?

[jc101]

There is a group who keep an Infrastructure map, like open street maps, but for services which might be of interest to some...

https://openinframap.org

[AlbertL]

The most common distribution system, called a "radial" system, consists of a feeder circuit (generally in the 4-35 kV range) originating at a substation and supplying multiple transformers, each of which serves one or more customers.  The key characteristic of this system is that any given customer is supplied by only one feeder circuit.

In the UK we have distribution circuits like this at 11 and 33 kV (phase-phase), but ours tend to be a bit different. In most cases the feeder forms a ring from the primary substation, around some number (perhaps 10-15) secondary substations each with 1 or 2 transformers supplying a few hundred customers each, and back to the primary substation. The ring is broken at one point, and is fed from both ends essentially creating two radial circuits. The nice feature though is that the break can be remotely switched to occur at any of the secondary substations, or at two of them, such that any one faulty length of cable can be isolated for repair without loosing the 11 kV feed to any of the secondary substations.

My utility (Dominion Virginia Power) has something similar in their overhead systems.  Single or two-circuit high-voltage (230kV I think) transmission lines feed large substations, which step down to 34kV feeders.  These feeders either supply distribution transformers directly or (in older neighborhoods) supply smaller substations which step down to 12.5kV feeders for final distribution.  Where a feeder from one substation is routed near a feeder from another, a "tie switch" is installed so that, if a feeder fails, the entire load can be served by one substation after the other end of the feeder is disconnected.  A lineman told me that, when a feeder has to be opened for maintenance, they sometimes close the tie switch while both feeders are energized, thus paralleling the transformers at the two substations, to avoid a service interruption.         

In a secondary network, customers are served by two or more transformers paralleled on the secondary side and supplied by different feeder circuits, possibly from different substations.  The simplest form is the "spot" network, having only one customer on the secondary bus.  In an "area" network, the bus serves multiple customers.

We do have substations with multiple transformers and a common busbar, but so far as I am aware it's more common to have the busbar split into sections rather than operate transformers in parallel in normal operation. We commonly have LV feeders where each end connects to two secondary substations, which may or may not be on the same 11 kV feed. Most often these are connected as radial feeders, fed from one end only or fed from both with a break in the middle, but in areas with very high load density (central London for instance) the network is sometimes operated meshed with two or more substations feeding in. This gives rise to very high levels of fault current, and exciting failure modes if HV is lost at one site but not another. It gets interesting when a sizeable area ends up with its 11 kV supply powered via a distribution transformer operating in reverse and a bit of very hot-running LV cable down the street from the next substation.

Yes, that's the problem network protectors are intended to prevent.  I've read about incidents where the only way to disconnect the supply was to send linemen into manholes with hacksaws to cut live LV cables.  BTW, there's a type of fuse called a "cable limiter", which is rated by conductor size rather than amperage, that's made expressly for the protection of LV network cable insulation against thermal damage.   

A key component of a secondary network is the network protector.  This is a form of circuit breaker inserted between the secondary of each transformer and the bus.  Its purpose is to protect the system in case of a short circuit ("fault") or outage on a feeder by preventing reverse power flow from the secondary bus through the transformer into the primary feeder.     

I don't have a huge amount of experience with UK meshed networks (only a few towns have them), but as I understand it in most cases this protection is not provided and reverse power flow is allowed to persist until remote network reconfiguration can be used to restore power to the faulted section or until simple overload protection operates. Withstanding short-term operation like this is part of the design.

That's an interesting approach - potentially better if there's good (fast) telemetry in place.  My understanding of US practice is that the feeder breakers are not set low enough to operate on secondary faults, so they won't trip a feeder that's supplying a fault through a backfed transformer.

For example, in a simple spot network with Feeder A supplying Transformer A, and Feeder B supplying Transformer B, a fault on Feeder A will operate the Feeder A breaker at the substation, but the fault will still be powered by Feeder B through Transformer B and (backfed) Transformer A, and the current on Feeder B will not be high enough (due to the impedances of the two transformers back-to-back) to trip the Feeder B breaker.  So, without the network protector to disconnect Transformer A from the bus, the fault will continue to burn.     

[richard.cs]

For example, in a simple spot network with Feeder A supplying Transformer A, and Feeder B supplying Transformer B, a fault on Feeder A will operate the Feeder A breaker at the substation, but the fault will still be powered by Feeder B through Transformer B and (backfed) Transformer A, and the current on Feeder B will not be high enough (due to the impedances of the two transformers back-to-back) to trip the Feeder B breaker.  So, without the network protector to disconnect Transformer A from the bus, the fault will continue to burn.   

Here the expectation would be that the forward protection on the LV side of transformer B, either an oil circuit breaker or a fuse, would then operate, as it sees both the transformed feeder A fault current and its normal load current, plus quite possibly the load current of any other substations connected to feeder A. More commonly here Transformers A and B would not be co-located, but might feed opposite ends of perhaps 200m of cable with various loads along it, protected each end by a fuse in the region of 400 A. Transformers A and B might have other transformers in the same substation, but often feeding different cables, and perhaps meshed with different substations, from other sections of a split LV board.

A typical UK arrangement in a high-load area might have two transformers of perhaps 1 MVA in each substation, feeding two sides of a split LV distribution board with a normally-open switch in the middle of the busbars. There may be an oil circuit breaker on the LV side of the transformer before the board, but commonly there isn't and the only protection for those cables and busbars is on the HV side. From each half of the board cables go out in many directions, most with various branches, and loads tapped off them along their length. Typically these are 3 phase cables of around 95-400 mm² protected by fuses in the range 300-500A, perhaps something in the region of 20 such feeders. Many of those cables connect to underground linkboxes where fuses or most-often solid links can be fitted, and cables from other substations also come to the same boxes. A given length of cable and its various loads can then be fed from either end, or from both with a break at some point (not necessarily in the centre), or from two or more places in areas with very dense loads (referred to as meshed operation).

In most of the UK loads are generally fed radially from one substation but exactly which can be changed by physically moving links/fuses, which is often done during maintenance to limit the number of customers off-supply, or to manually re-configure to link out faulted sections or even whole substations. It is also done occasionally to move loads from a more to a less loaded substation, and occasionally both substation LV boards and underground linkboxes are used as convinient points to connect diesel generators. Generally links are fitted and removed live, with substations paralleled for a small number of minutes to avoid loosing power to the section that's being reconfigured. In some urban areas (possibly only central London and Liverpool) the LV network is operated meshed, and many substations are connected together at LV via these cables and linkboxes, giving rise to the backfeeding faults and in some cases fairly specialised protection is in place to limit their duration or improve fault clearance times.

In a low load area a substation might have a single transformer of 250-500 kVA, and single, small LV board with perhaps 4-6 outgoing ways. The LV cabling is still likely to have a mesh-like structure, but generally with open points at linkboxes.

[AlbertL]

This might be of interest to give readers an idea of what we're discussing - it's a brochure for the Eaton CM52 network protector, which I believe is one of the most commonly-used models in the US: https://www.eaton.com/ecm/idcplg?IdcService=GET_FILE&allowInterrupt=1&RevisionSelectionMethod=LatestReleased&noSaveAs=0&Rendition=Primary&dDocName=CA024006EN

[richard.cs]

And some examples of the kind of thing I was talking about in the UK:
Typical mid-size (maybe 500 KVA?) urban substation, probably 1950s, open frame LV board (plenty like this still in service) https://www.geograph.org.uk/photo/2847151
Typical suburban substation, outdoor transformer and LV distribution in a metal cabinet, perhaps 250 kVA at most, perhaps 1970s-1990s https://commons.wikimedia.org/wiki/File:Cat_napping_on_substation_Portsmouth_UK_2018.jpg
1960s LV board from a larger (but still 11kV/415V) substation with two transformers https://www.reddit.com/r/electricians/comments/f9s2el/dont_know_if_this_will_interest_any_of_you_but/

I have some better photos on my work PC, but it's Sunday at the moment so that was 5 mins random googling. A lot of the modern ones just look like big grey cabinets with nothing to really see, and there's really not a lot of good photos online (expected as there is as no public access).

About the only reasonable photo I could find of a UKish* primary substation (transmission voltage down to 11 or 33 kV) are here: https://www.jec.co.uk/about-us/latest-news/%C2%A317m-new-primary-substation-is-successfully-%E2%80%98switched-on%E2%80%99/

* Technically not UK as it's Jersey, but it looks very similar to the UK ones.

[AlbertL]

A key component of a secondary network is the network protector.  This is a form of circuit breaker inserted between the secondary of each transformer and the bus.  Its purpose is to protect the system in case of a short circuit ("fault") or outage on a feeder by preventing reverse power flow from the secondary bus through the transformer into the primary feeder.     

How does that work when there's solar or other grid tie energy sources on the customer side? Is the reverse power breaker set to trip only if the backflow is greater than the sum of all distributed energy generation in that zone?

In fact, that's the reason I got involved with secondary networks.  I've worked for several solar companies in Washington, DC who have installed or want to install grid-tied PV systems on buildings served by secondary networks.  The utility (Pepco) requires protective relaying on such systems; specifically, a minimum import relay on spot networks, and a maximum export relay on area networks.

Right now I'm working on a project that will put about 70 kW DC (module nameplate capacity) / 57 kW AC (inverter nameplate capacity) on a building served by a spot network.  The utility has specified a 20 kW minimum import.  Over the past year, the building had a maximum load (or "demand"; the average over a 15-minute interval) of 140 kW and a minimum of 38 kW, so with the solar there's a substantial likelihood of pushing the demand below 20 kW.  Furthermore, the building owner plans substantial HVAC energy-efficiency upgrades.

The relay requirement is pretty stringent as regards timing: the relay must operate before the network protector (which makes sense of course).  I'm having trouble finding out what the protector's "pickup" time setting is, but I've heard it's 9 cycles (150 ms).  If by "operate" the utility actually means "clear", then we have to add the opening time of the circuit breaker (or response time of the inverters' external shutdown feature) to our relay's pickup time.

Because of the likelihood of minimum-import tripping, we'll install a "curtailment" system that will ramp down the inverter output before net power flow reaches the 20-kW minimum.  This has its own set of problems.  Solar in DC is a lucrative investment due to a production incentive, currently worth about $0.40/kWh, on top of the purchased electricity savings of about $0.11/kWh, so we don't want to lose any more production than necessary.  At the same time, a curtailment algorithm that's not aggressive enough will result in relay tripping that may require manual intervention to reset.       

[richard.cs]

Ah yes, the joy of trying to make systems that aren't designed to work together play nicely. It seems like the "protector" settings are a bit on the frisky side, and having a non-zero minimum forward power just seems odd. Am I reading that right that this building must draw >20 kW at all times to not get disconnected?

I have had some involvement with similar protection devices here in the UK, where they are not especially common (as the LV meshes are uncommon, and most of the ones that exist don't have such protection), and a more typical approach here would be to trip in 100-200 ms at reverse power levels on the order of -50 to -200% of the substation nominal rating (500-1000 kVA generally). The idea being to avoid excessively high fault-clearance times on the 11 kV side due to backfeed. The reverse power due to a real HV side fault is never going to be small after all. In some cases this is combined with voltage information to further avoid reverse trips caused by export. Reverse-power into a HV-fault would correspond with low LV voltage whereas reverse power due to export will correspond to high HV voltages so the two conditions can be distinguished.

[AlbertL]

As regards the 20 kW minimum, the wording of the utility's requirement doesn't explicitly limit it to times when the PV system is producing, but since it requires curtailing or tripping the generation in case of a violation, that appears to be the intent:

Install controls and relay(s) that monitor the entire load of the spot network on all three phases. The control shall be set up to maintain a minimum power import of 20 kW. If import falls below 20 kW, the control system shall curtail or trip the generation. When this limit is exceeded, response time for curtailment or trip should occur instantaneously.

Which of course brings up the question: what bad thing would happen if import falls below 20 kW at night, due to the building simply not consuming much power?

A utility engineer did tell me that the reason for the minimum import (as opposed to just zero export) is that if the voltage on one feeder is lower than the other, and there's no load on the collector bus, there can be enough reverse power flow through that feeder's transformer to trip the protector.  The minimum import ensures that there's always a forward power flow through the protectors under all normal feeder voltage variations.  But again, a feeder voltage differential can just as easily occur at night.  I think in the end it comes down to playing the odds.

 

[NiHaoMike]

Right now I'm working on a project that will put about 70 kW DC (module nameplate capacity) / 57 kW AC (inverter nameplate capacity) on a building served by a spot network.  The utility has specified a 20 kW minimum import.  Over the past year, the building had a maximum load (or "demand"; the average over a 15-minute interval) of 140 kW and a minimum of 38 kW, so with the solar there's a substantial likelihood of pushing the demand below 20 kW.  Furthermore, the building owner plans substantial HVAC energy-efficiency upgrades.

The relay requirement is pretty stringent as regards timing: the relay must operate before the network protector (which makes sense of course).  I'm having trouble finding out what the protector's "pickup" time setting is, but I've heard it's 9 cycles (150 ms).  If by "operate" the utility actually means "clear", then we have to add the opening time of the circuit breaker (or response time of the inverters' external shutdown feature) to our relay's pickup time.

Because of the likelihood of minimum-import tripping, we'll install a "curtailment" system that will ramp down the inverter output before net power flow reaches the 20-kW minimum.  This has its own set of problems.  Solar in DC is a lucrative investment due to a production incentive, currently worth about $0.40/kWh, on top of the purchased electricity savings of about $0.11/kWh, so we don't want to lose any more production than necessary.  At the same time, a curtailment algorithm that's not aggressive enough will result in relay tripping that may require manual intervention to reset.       

What is the majority of the load? Is there a significant amount that would make sense to try to schedule to run more when solar is available? For example, dedicated freezers or water heaters? HVAC can also be controlled to a more limited extent, perhaps by increasing the fresh air percentage.

Could also propose the idea of renting out a little space for cryptocurrency mining or other computing.

[richard.cs]

A utility engineer did tell me that the reason for the minimum import (as opposed to just zero export) is that if the voltage on one feeder is lower than the other, and there's no load on the collector bus, there can be enough reverse power flow through that feeder's transformer to trip the protector.  The minimum import ensures that there's always a forward power flow through the protectors under all normal feeder voltage variations.  But again, a feeder voltage differential can just as easily occur at night.  I think in the end it comes down to playing the odds.

Yes, it sounds like they have set an unreasonable "zero reverse allowable, fast trip" setting and a minimum load is something of an ugly workaround. It's not really clear why they have done so, but I guess you're stuck with it. Actually it hints at a bit of a wider issue, the supplier responsibility would appear to end at the output of each separate supply, and the parallel connection is the responsibility of the building owner who then has to work around protection settings outside of their control. Is that the norm in the US?

What is the majority of the load? Is there a significant amount that would make sense to try to schedule to run more when solar is available? For example, dedicated freezers or water heaters? HVAC can also be controlled to a more limited extent, perhaps by increasing the fresh air percentage.

Could also propose the idea of renting out a little space for cryptocurrency mining or other computing.

This sounds like a reasonable workaround to me. Do something useful with the energy you're not allowed to export and also helps meet the minimum load.

[AlbertL]

Right now I'm working on a project that will put about 70 kW DC (module nameplate capacity) / 57 kW AC (inverter nameplate capacity) on a building served by a spot network.  The utility has specified a 20 kW minimum import.  Over the past year, the building had a maximum load (or "demand"; the average over a 15-minute interval) of 140 kW and a minimum of 38 kW, so with the solar there's a substantial likelihood of pushing the demand below 20 kW.  Furthermore, the building owner plans substantial HVAC energy-efficiency upgrades.

The relay requirement is pretty stringent as regards timing: the relay must operate before the network protector (which makes sense of course).  I'm having trouble finding out what the protector's "pickup" time setting is, but I've heard it's 9 cycles (150 ms).  If by "operate" the utility actually means "clear", then we have to add the opening time of the circuit breaker (or response time of the inverters' external shutdown feature) to our relay's pickup time.

Because of the likelihood of minimum-import tripping, we'll install a "curtailment" system that will ramp down the inverter output before net power flow reaches the 20-kW minimum.  This has its own set of problems.  Solar in DC is a lucrative investment due to a production incentive, currently worth about $0.40/kWh, on top of the purchased electricity savings of about $0.11/kWh, so we don't want to lose any more production than necessary.  At the same time, a curtailment algorithm that's not aggressive enough will result in relay tripping that may require manual intervention to reset.       

What is the majority of the load? Is there a significant amount that would make sense to try to schedule to run more when solar is available? For example, dedicated freezers or water heaters? HVAC can also be controlled to a more limited extent, perhaps by increasing the fresh air percentage.

Could also propose the idea of renting out a little space for cryptocurrency mining or other computing.

Those are good ideas.  I had proposed something cruder: a resistive load bank, normally used for testing generators and other power sources.  As far as the production incentive is concerned, it doesn't matter how the energy is used (or wasted) as long as it's delivered to a building served by the utility. 

[Red Squirrel]

There is a group who keep an Infrastructure map, like open street maps, but for services which might be of interest to some...

https://openinframap.org

Wow that's really cool. I always nerd out at power transmission lines and wonder where they go or come from and sometimes even check google maps. I had no idea we had a 500kv one in my area but I found it when I wanted to get pictures of it and realized it was literally raising my arm hairs.  I knew it had to be higher voltage than the other lines I've been around so got on google maps to follow it to the hydro dam and all the way to the GTA.   Seeing this map is cool as it does confirm my findings.

And if curious this is the 500kv line, I thought the pic turned out pretty cool. 

https:// i.imgur.com/6VxxQ3K.jpg

(need to manually copy and paste url, forum keeps breaking it)

I need to go back there one night with some neon tubes to see if they light up.   

[SG-1]

  The relay that controls the breaker requires an external load to be present before closing.  On the CMD Network Protector there was an option that could be purchased called a phantom load.  The relay that controls the breaker would see an external load when none actually existed.  Since the utility owns the protectors, that option is most likely off the table.

We would test the protector by attaching an external load, like a light bulb & make sure the protector would close.

The network protector product line was Westinghouse's oldest product line when sold to Eaton.  It has been in production since the late 1800s.