Power Transformers In Parallel

[mbrooke]

It depends on a number of things:
- system configuration (transformers in parallel, radial feed, or secondary selective system)
- primary side system stiffness - if we already have fairly low fault current, we probably don’t want to knock it down further with a high impedance transformer
- what kind of load you’re feeding (steady, cyclical, any big motors to start)
- downstream distribution system characteristics - what additional voltage levels are involved, how large is the distribution system, what losses are expected downstream
- power price per kWh to pay for transformer losses
- voltage regulation on the primary
- what voltage regulation is required on the secondary.

Without assessing all of those details, the short answer is 8% sounds high. It might allow you to use 31 kA MV gear, but the voltage regulation and efficiency benefits may justify 6-7% Z even though the switchgear may need to be rated for more fault current.


Sent from my iPhone using Tapatalk

Primary 115kv is allowed to vary + or - 5%.

Voltage on the secondary is regulated at +5% during peak periods, sometimes a tad higher under extreme contingencies.


I'm curious how radial overhead vs network is effected by a low Z bank vs a high Z bank.


FWIW, normally open secondary 20/33MVA, 30/40/50MVA and 40/50/60MVA is serving loads in POCO right now satisfactorily. This is the base concpet I am comparing everything to.

 

[jdsmith]

Primary 115kv is allowed to vary + or - 5%.

Voltage on the secondary is regulated at +5% during peak periods, sometimes a tad higher under extreme contingencies.


I'm curious how radial overhead vs network is effected by a low Z bank vs a high Z bank.


FWIW, normally open secondary 20/33MVA, 30/40/50MVA and 40/50/60MVA is serving loads in POCO right now satisfactorily. This is the base concpet I am comparing everything to.

That primary voltage tolerance is pretty typical. How is the secondary voltage regulated - will these new transformers be equipped with on-load tap changers (OLTC)?

I didn’t see it explicitly stated in this thread, but it sounds like we are assessing a utility substation where we have minimal context into the load details, rather than a commercial/industrial site where we have a detailed understanding of the load. If this assumption is correct it also means we aren’t concerned with large motor starting.

If this is a utility system the fault current constraints are quite different - we have to keep the fault currents within the limits of the reclosers and cutouts. Assuming a cutout on the first pole outside the substation, now we’re edging back toward higher impedance transformers again. This keeps fault current within those recloser and cutout ratings but makes voltage regulation worse. An OLTC is the best solution here, but at a fairly high cost. We also have to worry about voltage regulation at the end of the distribution circuit, which includes the effects of distribution line impedance and how loads are distributed along the line. This analysis exercise is straightforward but it does require a fair amount of data to obtain an acceptable outcome.

Now consider the above concerns on a 23 kV system where the distribution lines are fed from both ends (this being the simplest example of a network system). We now have multiple sources feeding any fault and we still have to maintain short circuit levels within the same recloser and cutout equipment ratings. Voltage regulation behaves differently since the circuit is fed from both ends and there are chunks of load and pieces of impedance intermingled throughout the network. Developing rules of thumb here is more difficult given the level of interconnection and how many different impedance elements are involved in the network. Assessing optimal transformer size and impedance in a networked system requires software-based load flow and short circuit study runs to test different equipment options, different load cases, and different normal and abnormal operating conditions.


Sent from my iPhone using Tapatalk

 

[mbrooke]

That primary voltage tolerance is pretty typical. How is the secondary voltage regulated - will these new transformers be equipped with on-load tap changers (OLTC)?

OLTC, unless there is compelling benefit from individual feeder regulation.

I didn’t see it explicitly stated in this thread, but it sounds like we are assessing a utility substation where we have minimal context into the load details, rather than a commercial/industrial site where we have a detailed understanding of the load. If this assumption is correct it also means we aren’t concerned with large motor starting.

Correct, utility substation. A mix of residential, commercial, small industrial and institutional loads.

If this is a utility system the fault current constraints are quite different - we have to keep the fault currents within the limits of the reclosers and cutouts. Assuming a cutout on the first pole outside the substation, now we’re edging back toward higher impedance transformers again. This keeps fault current within those recloser and cutout ratings but makes voltage regulation worse. An OLTC is the best solution here, but at a fairly high cost. We also have to worry about voltage regulation at the end of the distribution circuit, which includes the effects of distribution line impedance and how loads are distributed along the line. This analysis exercise is straightforward but it does require a fair amount of data to obtain an acceptable outcome.

Correct, primary fault current can vary between 5,000 amps to 63,000 amps depending on where in the system.


While expulsion cutouts are rated 12,000 amps, things like back-up companion II fuses and ELF drop outs can increase the rating.

Now consider the above concerns on a 23 kV system where the distribution lines are fed from both ends (this being the simplest example of a network system). We now have multiple sources feeding any fault and we still have to maintain short circuit levels within the same recloser and cutout equipment ratings. Voltage regulation behaves differently since the circuit is fed from both ends and there are chunks of load and pieces of impedance intermingled throughout the network. Developing rules of thumb here is more difficult given the level of interconnection and how many different impedance elements are involved in the network. Assessing optimal transformer size and impedance in a networked system requires software-based load flow and short circuit study runs to test different equipment options, different load cases, and different normal and abnormal operating conditions.

Understood. In this case, all ties are run normally open.

 

[mbrooke]

That primary voltage tolerance is pretty typical. How is the secondary voltage regulated - will these new transformers be equipped with on-load tap changers (OLTC)?

I didn’t see it explicitly stated in this thread, but it sounds like we are assessing a utility substation where we have minimal context into the load details, rather than a commercial/industrial site where we have a detailed understanding of the load. If this assumption is correct it also means we aren’t concerned with large motor starting.

If this is a utility system the fault current constraints are quite different - we have to keep the fault currents within the limits of the reclosers and cutouts. Assuming a cutout on the first pole outside the substation, now we’re edging back toward higher impedance transformers again. This keeps fault current within those recloser and cutout ratings but makes voltage regulation worse. An OLTC is the best solution here, but at a fairly high cost. We also have to worry about voltage regulation at the end of the distribution circuit, which includes the effects of distribution line impedance and how loads are distributed along the line. This analysis exercise is straightforward but it does require a fair amount of data to obtain an acceptable outcome.

Now consider the above concerns on a 23 kV system where the distribution lines are fed from both ends (this being the simplest example of a network system). We now have multiple sources feeding any fault and we still have to maintain short circuit levels within the same recloser and cutout equipment ratings. Voltage regulation behaves differently since the circuit is fed from both ends and there are chunks of load and pieces of impedance intermingled throughout the network. Developing rules of thumb here is more difficult given the level of interconnection and how many different impedance elements are involved in the network. Assessing optimal transformer size and impedance in a networked system requires software-based load flow and short circuit study runs to test different equipment options, different load cases, and different normal and abnormal operating conditions.


Sent from my iPhone using Tapatalk

One advantage I see in the parallel scheme is that the loss of one unit does not drop customers, where as in the MTM scheme the loss of a unit momentarily drops 12,000+ customers, in addition to having to pickup hot load inrush when closing the bus tie there after... I wonder if so many stalled AC units will not lead to voltage collapse on the remaining secondary with rapid closing...

Of course both 60MVA units could be operated in parallel, but assuming a strong transmission source that would result in 25,000 amps of 3 phase fault current vs 15,000 amps of 3 phase fault current on the secondary.

A strong transmission source example would be distro coming from a 500kv/230kv/23kv substation- 3 600MVA or 4 400MVA autos in parellel makes for a stiff Z.

 

[jdsmith]

One advantage I see in the parallel scheme is that the loss of one unit does not drop customers, where as in the MTM scheme the loss of a unit momentarily drops 12,000+ customers, in addition to having to pickup hot load inrush when closing the bus tie there after... I wonder if so many stalled AC units will not lead to voltage collapse on the remaining secondary with rapid closing...

Of course both 60MVA units could be operated in parallel, but assuming a strong transmission source that would result in 25,000 amps of 3 phase fault current vs 15,000 amps of 3 phase fault current on the secondary.

A strong transmission source example would be distro coming from a 500kv/230kv/23kv substation- 3 600MVA or 4 400MVA autos in parellel makes for a stiff Z.

That is the main advantage, and that’s why some industrial systems run closed tie. To mitigate the fault current with a MTM configuration we may use an air core reactor on the tie, or we use switchgear that is rated to withstand the full fault current. There are numerous other common industrial topologies involving reactors and tie buses. Trying to find an engineer that can conduct the stability studies to correctly size the tie reactor is not as easy as developing the topology

To directly address the crux of your question, these additional reliability measures are usually not economical justified for a utility based on SAIDI, SAIFI, and the other reliability metrics utilities use. These other systems you propose are fun to work on, but they’re relegated to the continuous industrial process industry sector - places like oil refineries, chemical plants, paper mills, and semiconductor manufacturing. Finding those facility owners who are willing to spend the capital to build a highly reliable power system is becoming more difficult though.


Sent from my iPhone using Tapatalk

 

[mbrooke]

That is the main advantage, and that’s why some industrial systems run closed tie. To mitigate the fault current with a MTM configuration we may use an air core reactor on the tie, or we use switchgear that is rated to withstand the full fault current. There are numerous other common industrial topologies involving reactors and tie buses. Trying to find an engineer that can conduct the stability studies to correctly size the tie reactor is not as easy as developing the topology

To directly address the crux of your question, these additional reliability measures are usually not economical justified for a utility based on SAIDI, SAIFI, and the other reliability metrics utilities use. These other systems you propose are fun to work on, but they’re relegated to the continuous industrial process industry sector - places like oil refineries, chemical plants, paper mills, and semiconductor manufacturing. Finding those facility owners who are willing to spend the capital to build a highly reliable power system is becoming more difficult though.


Sent from my iPhone using Tapatalk

Ehhh, that is until you see Con Edison's system or PSE&G's 26.4kv, mostly run normally closed.


But I do get you- 260MW of networked load in Queens/Brooklyn or sub-transmission are in a higher reliability matrix.

With that said, I'm starting to think that POCOs have evaluated multiple smaller units vs fewer larger units and have found the latter cheaper?

 

[jdsmith]

Ehhh, that is until you see Con Edison's system or PSE&G's 26.4kv, mostly run normally closed.


But I do get you- 260MW of networked load in Queens/Brooklyn or sub-transmission are in a higher reliability matrix.

With that said, I'm starting to think that POCOs have evaluated multiple smaller units vs fewer larger units and have found the latter cheaper?

Like protective relaying, I think power system design philosophies are also and art and a science. I haven’t studied higher voltage/density distribution systems in detail - there may be some technical reason those two utilities are using networked systems at that voltage. Or it may come down to the particular accounting rules that govern the regulated utility sector where profit is a function of total caped and opex, so in some cases utilities are incentivized to spend as much money as they can get away with.

I know the industrial sector has evaluated multiple smaller vs fewer larger transformers and concluded fewer transformers usually presents a lower total installed cost per MVA of transformation. While there are numerous differences between utility and industrial power systems, I suspect the major utilities have also evaluated this and come to the same conclusion.


Sent from my iPhone using Tapatalk

 

[mbrooke]

Like protective relaying, I think power system design philosophies are also and art and a science. I haven’t studied higher voltage/density distribution systems in detail - there may be some technical reason those two utilities are using networked systems at that voltage. Or it may come down to the particular accounting rules that govern the regulated utility sector where profit is a function of total caped and opex, so in some cases utilities are incentivized to spend as much money as they can get away with.

I know the industrial sector has evaluated multiple smaller vs fewer larger transformers and concluded fewer transformers usually presents a lower total installed cost per MVA of transformation. While there are numerous differences between utility and industrial power systems, I suspect the major utilities have also evaluated this and come to the same conclusion.


Sent from my iPhone using Tapatalk

I'd say so, most likely fewer larger units are less cost.

There is also the fact more primary protective equipment is needed- so it indeed might be less cost per MW.

 

[NewtonLaw]

The AIC is the “rating” of fault current an interrupting device can safely interrupt.

A higher system X/R results in a higher magnitude fault duty.

A higher X/R ratio will have little affect on the symmetrical AIC but a very larger affect on the asymmetrical levels as you get closer to the supply transformer. It will not matter if it is a single transformer or multiple transformers as long as you the equivalent MVA and %Z and X/R ratio. What will affect the asymmetrical levels is the length of the supply conductor to the point of fault. The longer the supply conductors to the point of fault, the closer the symmetrical and asymmetrical values become. I held the X/R ratios to either 5 or 40 to compare the output short circuit currents with all transformers having an 8%Z with matching voltage ratings.

Having said this however, I note that the normal X/R ratios of a single large transformer is almost always higher than that of a smaller unit. A 60 MVA unit would typically be 35 to 40 X/R while a 10 MVA unit would be closer to 16 for an X/R ratio.

 

[mbrooke]

A higher X/R ratio will have little affect on the symmetrical AIC but a very larger affect on the asymmetrical levels as you get closer to the supply transformer. It will not matter if it is a single transformer or multiple transformers as long as you the equivalent MVA and %Z and X/R ratio. What will affect the asymmetrical levels is the length of the supply conductor to the point of fault. The longer the supply conductors to the point of fault, the closer the symmetrical and asymmetrical values become. I held the X/R ratios to either 5 or 40 to compare the output short circuit currents with all transformers having an 8%Z with matching voltage ratings.

Having said this however, I note that the normal X/R ratios of a single large transformer is almost always higher than that of a smaller unit. A 60 MVA unit would typically be 35 to 40 X/R while a 10 MVA unit would be closer to 16 for an X/R ratio.

That was my hunch.

I feel as though smaller units are more "optimal" while larger units are a bit "frankenstein"

When you say the symmetrical and asymmetrical values become closer as a function of conductor distance, what do you mean by this? Does conductor size matter, particularly larger sizes where X dominates over R?

 

[jdsmith]

That was my hunch.

I feel as though smaller units are more "optimal" while larger units are a bit "frankenstein"

When you say the symmetrical and asymmetrical values become closer as a function of conductor distance, what do you mean by this? Does conductor size matter, particularly larger sizes where X dominates over R?

Lol, ‘optimal’ and ‘Frankenstein’ don’t mean anything as long as the asymmetrical fault current at the cutout is within its rating. (You may choose to nitpick my terminology here but I chose it carefully. Fuses have a symmetrical fault current rating up to a specified X/R value, for X/R above the specified value the symmetrical rating must be derated. So it’s faster to refer to fuses as having a particular asymmetrical rating even though that’s not technically how they are rated and tested, it still reflects the physics of how they work.)

Assuming we’re talking about overhead distribution that has a lot of X per mile and small-medium R per mile, it doesn’t take very much OH line to counteract the higher X/R of the larger transformer. If we ran a short circuit report a fairly small distance down the distribution circuit the symmetrical/asymmetrical fault currents would be close to the same whether we were fed from several small transformers in parallel or the one large transformer.

Conductor size of an overhead circuit changes R. The conductor spacing determines the X. It is reasonable to approximate X as being constant for a particular pole line design and spacing as the conductor size changes. The subtlety is that most overhead line configurations do not have the phases arranged symmetrically, so the mutual inductance between phases will be different, which tends to create or accentuate voltage unbalance. On long transmission circuits phase conductors may be transposed to minimize this issue.


Sent from my iPhone using Tapatalk

 

[xptpcrewx]

A higher X/R ratio will have little affect on the symmetrical AIC but a very larger affect on the asymmetrical levels as you get closer to the supply transformer. It will not matter if it is a single transformer or multiple transformers as long as you the equivalent MVA and %Z and X/R ratio.

Not sure what you’re getting at. No one is saying X/R is particularly important for the 30 cycle calculation. The comment was simply about how oversimplifying by using the 30 cycle (symmetrical) fault current for determining the fault duty is incorrect and an overall bad assumption to make.

As far as paralleling units of the same type, the fault current doubles with the X/R staying relatively the same. This results in double the asymmetrical fault current as compared to one unit (nothing surprising there). The fact you are paralleling units changes the effective MVA and %Z. The OP is not considering the asymmetrical factor.

 

[mbrooke]

The OP is not considering the asymmetrical factor.

Still good to know/learn. Soon or latter asymmetrical factor will be called into consideration as will everything else from the soil PH on the ground grid to the cyber security on the SCADA.

 

[xptpcrewx]

Still good to know/learn. Soon or latter asymmetrical factor will be called into consideration as will everything else from the soil PH on the ground grid to the cyber security on the SCADA.

Asymmetrical factor has always been a consideration in applying protective devices. You can lean more about this from the IEEE color book series.

 

[mbrooke]

Asymmetrical factor has always been a consideration in applying protective devices. You can lean more about this from the IEEE color book series.

Of course!

IEEE Greenbook even tells use that 250.122 can be inadequate.

 

[mbrooke]

Lol, ‘optimal’ and ‘Frankenstein’ don’t mean anything as long as the asymmetrical fault current at the cutout is within its rating. (You may choose to nitpick my terminology here but I chose it carefully. Fuses have a symmetrical fault current rating up to a specified X/R value, for X/R above the specified value the symmetrical rating must be derated. So it’s faster to refer to fuses as having a particular asymmetrical rating even though that’s not technically how they are rated and tested, it still reflects the physics of how they work.)

Assuming we’re talking about overhead distribution that has a lot of X per mile and small-medium R per mile, it doesn’t take very much OH line to counteract the higher X/R of the larger transformer. If we ran a short circuit report a fairly small distance down the distribution circuit the symmetrical/asymmetrical fault currents would be close to the same whether we were fed from several small transformers in parallel or the one large transformer.

Conductor size of an overhead circuit changes R. The conductor spacing determines the X. It is reasonable to approximate X as being constant for a particular pole line design and spacing as the conductor size changes. The subtlety is that most overhead line configurations do not have the phases arranged symmetrically, so the mutual inductance between phases will be different, which tends to create or accentuate voltage unbalance. On long transmission circuits phase conductors may be transposed to minimize this issue.


Sent from my iPhone using Tapatalk

I know. But there is something that just feels "right" about smaller units, ie they are easier to transport vs shutting down streets to move larger ones.


Nahhh, I'm not going to nitpick, you are correct


Question- why does line X counteract the trafo X? Wouldn't the X or the trafo and the line be additive making X/R ratio higher?

 

[mbrooke]

Forgot to ask, how is voltage regulation? Better or worse?

 

[xptpcrewx]

Forgot to ask, how is voltage regulation? Better or worse?

If you know basic circuit theory you can easily calculate it yourself.

 

[mbrooke]

Thanks again for your time

 

[Mr. Serious]

I started reading this thread last month before there were so many replies, and one of the first questions you asked was

Any idea why POCOs always use two large transformers in a substation?

I didn't know the answer at the time, and it seemed like the original people replying also didn't know.

Since then, I've been putting some transmission lines into Open Street Map, and I've noticed that often a substation has two transmission lines feeding from different directions, with one transformer connected to each. Presumably these transformers are operated in parallel for reliability reasons if one transmission line goes down. And, it's simpler to use two transformers than multiple smaller ones.

So, that would seem to be the simple answer. which I think has by now been strongly implied by the additional responses since the last time I was here.

It is also correct to say

Not all POCO substations have two transformers

It really depends on the common practice for the utility and area you're working in. In rural areas, there can be just one transformer in each substation, a pretty small one depending on the load served.