3 Phase Neutral Reduction

[JDye]

Hello and good evening.

I am trying to size a service into a commercial/retail setting and am a little confused on the wording of the NEC code book. I'm trying to size my neutral in a 2500 amp 3 phase 120/208 setting. The wires in question will run about 50 feet from the transformer to the main disconnect. The plan is to run 6 pipes of 600 KCMIL copper for the legs and since this is a large project, we are wondering if we can save the customer some money by running something like a 500 KCMIL (or smaller but only safely) copper for the neutrals.

I've read over and over the 2002, 220.22 (what the AHJ uses) and the 2017 220.61 A-C. And I've tried searching the forums but I didn't find anything that answered the type of question I had.

Maybe I'm just missing something here but first off, can I reduce the neutral? If so can I use the 70% or do I need to figure some other percentage?

I hope this isn't too much of newbie question and thank you all for the info.

 

[George Stolz]

I was scared that this was going to be a foodie thread.

What is you connected neutral load? What is the nature of the loads being served by the service?

 

[electrofelon]

In my experience it is very common, and I would almost say something like "industry standard" to size three phase service neutrals full size, and single phase neutrals two size reduced, however according to the NEC, it all comes down to what the load calc says.

 

[Besoeker]

What is you connected neutral load? What is the nature of the loads being served by the service?

Commercial/retail suggests that there could be a lot of non-linear, single-phase loads resulting in triple-n harmonics which add in the neutral.
I have seen cases where the neutral has had to be oversized compared to the line conductors.
And one case where it wasn't. That was a very expensive error.

 

[Besoeker]

Hello and good evening.

I am trying to size a service into a commercial/retail setting and am a little confused on the wording of the NEC code book. I'm trying to size my neutral in a 2500 amp 3 phase 120/208 setting. The wires in question will run about 50 feet from the transformer to the main disconnect. The plan is to run 6 pipes of 600 KCMIL copper for the legs and since this is a large project, we are wondering if we can save the customer some money by running something like a 500 KCMIL (or smaller but only safely) copper for the neutrals.

I've read over and over the 2002, 220.22 (what the AHJ uses) and the 2017 220.61 A-C. And I've tried searching the forums but I didn't find anything that answered the type of question I had.

Maybe I'm just missing something here but first off, can I reduce the neutral? If so can I use the 70% or do I need to figure some other percentage?

I hope this isn't too much of newbie question and thank you all for the info.

I wouldn't consider a reduced size neutral. See my reply to George Stolz.

 

[kwired]

Commercial/retail suggests that there could be a lot of non-linear, single-phase loads resulting in triple-n harmonics which add in the neutral.
I have seen cases where the neutral has had to be oversized compared to the line conductors.
And one case where it wasn't. That was a very expensive error.

A majority of the load needs to be non linear for this to be much of an issue.

Commercial/retail probably has a significant amount of load being HVAC - and even if HVAC is non linear (VFD driven compressor) it is likely line to line and not utilizing the neutral.

Something like a data center is one of the most popular applications where one may need to be more concerned about size of the neutral.

 

[kingpb]

Three questions:
1. Why?
2. Why?
3. Why?

Frankly, you will spend more time wondering, questioning, and toiling over it. How much are you really going to save today, that could end up be a dis-service later.

Run the neutral the same size and be done with it. No one will ever question it, or hate you later.

 

[Besoeker]

A majority of the load needs to be non linear for this to be much of an issue.

Commercial/retail probably has a significant amount of load being HVAC - and even if HVAC is non linear (VFD driven compressor) it is likely line to line and not utilizing the neutral.

Something like a data center is one of the most popular applications where one may need to be more concerned about size of the neutral.

The worst I came across was a hotel in Dubai. It wasn't even fully up and running before the neutral was cooking its bullocks off.

 

[JFletcher]

You're talking about the difference of 300 ft of 500 MCM copper versus 600mcm copper. I haven't priced copper lately, that might even be $1,000 difference, but like others have mentioned, it's not going to be worth it in the long run unless maybe perhaps the reduction in the neutral size let you get away with smaller conduit as well

 

[kingpb]

Answering from a contractor's perspective, if you already have the job, why take money out of your own pocket. If the difference is $1000, the OH&P mark-up on the $1000 is going to be at least $500. That's some nice beer money.

From an engineering perspective, I'm not saving the Owner anything by trying to cut it down to 500KCMIL because as the engineer I decide what is the best design, and if I cut it down then I now run the risk of looking really foolish when the inspector shows up and says, nope, gotta be full size. Now I've delayed the project $$ and the change order will be far greater than $1500.

If you want to look for cost savings, look at the difference between conduit and copper vs aluminum cable. You can run 7 x 750KCMIL aluminum instead of 6 x 600 KCMIL; conduit is still 4", this may be cheaper.

 

[kwired]

I had a project a couple years ago - updated to 1200 amp service - don't recall exactly what I ran for neutral - probably just went the typical 2 sizes smaller then ungrounded phase conductors. Same installation I had to run a new 600 amp feeder that supplied all new HVAC compressor units. Only two or three 120 volt circuits were supplied from that feeder lights and a receptacle or two in the compressor room. I ran whatever size neutral was allowed code wise (same as minimum EGC on a feeder) and that was still much larger then needed for the maximum current that neutral will ever see.

 

[JFletcher]

The worst I came across was a hotel in Dubai. It wasn't even fully up and running before the neutral was cooking its bullocks off.

How did that happen? Even if all the loads were phase to neutral, almost everything would have to be on just one or two phases to neutral to see as much or more current than the phase conductors.

Only realistic way I can see that happening outside of something like a data center is if you lost a phase from the power company, the neutral would then carry the entire unbalanced load of the other two phases. The possibility of that happening would encourage me to run a full size neutral on a 3-phase service

 

[kwired]

How did that happen? Even if all the loads were phase to neutral, almost everything would have to be on just one or two phases to neutral to see as much or more current than the phase conductors.

Only realistic way I can see that happening outside of something like a data center is if you lost a phase from the power company, the neutral would then carry the entire unbalanced load of the other two phases. The possibility of that happening would encourage me to run a full size neutral on a 3-phase service

But you would have one phase that isn't loaded either. If it were a warehouse, stadium, parking garage, or any other applicaiton where majority of load was lighting and you had loads mostly balanced - you would have about one third of the lighting not working - something that is likely noticed and not going to be left that way for very long.

If a place with three phase motors - single phasing motors are likely to shut down on overload or phase monitoring - again something that would get noticed.

 

[Besoeker]

How did that happen? Even if all the loads were phase to neutral, almost everything would have to be on just one or two phases to neutral to see as much or more current than the phase conductors.

Not so.
With all phases balanced with equal non-linear single phase loads the triple-n harmonics in all three add arithmetically in the neutral. Take the 3rd for example. Three times the supply frequency. So you get one complete cycle of it every 120 degrees of the three phase supply so end up in phase with each other. And add.

Only realistic way I can see that happening outside of something like a data center is if you lost a phase from the power company, the neutral would then carry the entire unbalanced load of the other two phases. The possibility of that happening would encourage me to run a full size neutral on a 3-phase service

You are not alone. It certainly caught our lighting division by surprise. As I have previously posted, my background was/is power electronics for medium to high power industrial applications so almost entirely 3-phase. Harmonics were always something to be considered - as early as the bid stage. Our lighting colleagues were blissfully unaware of this until it came to bite them on the backside. That's when their problems landed in my lap.

 

[electrofelon]

Those must have been real crappy light fixtures to have such a high harmonic content that the neutral got overloaded. Just measured a sub panel with 100 kW of electronic ballasts, balanced mwbc's. About 120 amps line current. Neutral current is about 30. Obviously can't make a sweeping generalizatin from one installation, but just sayin.

 

[Besoeker]

Those must have been real crappy light fixtures to have such a high harmonic content that the neutral got overloaded. Just measured a sub panel with 100 kW of electronic ballasts, balanced mwbc's. About 120 amps line current. Neutral current is about 30. Obviously can't make a sweeping generalizatin from one installation, but just sayin.

It wasn't a generalisation, sweeping or otherwise. It was a specific actual case. It happened. I was asked for my technical input on the problem.
Fourier series analysis was all it took* to see what was happening. And it wasn't crappy light fittings.
As I said, not everyone can get to grip with the issue.

*It's mathematics but mathematics is an integral part of engineering.

 

[electrofelon]

It wasn't a generalisation, sweeping or otherwise. It was a specific actual case. It happened. I was asked for my technical input on the problem.
Fourier series analysis was all it took* to see what was happening. And it wasn't crappy light fittings.
As I said, not everyone can get to grip with the issue.

*It's mathematics but mathematics is an integral part of engineering.

The comment about generalizations was in fact about my observations, and that I havnt seen this,issue in the field, therefore it doesn't exist

I have a BS in pure mathematics, I follow you.

So you don't think they were crappy light fixtures? IMO something with that high harmonic content is rather crappy, probably cheaply made, corners cut.

 

[Besoeker]

The comment about generalizations was in fact about my observations, and that I havnt seen this,issue in the field, therefore it doesn't exist

I have a BS in pure mathematics, I follow you.

Then you will obviously understand how the triple-n harmonics arithmetically sum in the neutral.

So you don't think they were crappy light fixtures? IMO something with that high harmonic content is rather crappy, probably cheaply made, corners cut.

Not so. This was a very high end luxury hotel* with professional consultants. Simply, the issue of harmonics was not understood.

*A grand a night to stay there, local transport was one of the world's largest fleet of Rolls Royce limos. Crappy it was not.

 

[electrofelon]

Sure, I understand the potential for harmonics adding in the neutral. What I am saying is it is very rare to have an electronic ballast/driver/power supply that results in such high neutral currents. I have never seen anything over 25% line current in the neuter. Been on this forum for years and this comes up regularly and never heard anyone say they have either. Perhaps in days past it was more of an issue before better topologies and PFC.

 

[Besoeker]

S

ure, I understand the potential for harmonics adding in the neutral.

It is not a potential. Triple-n add in the neutral. That is not in question.

What I am saying is it is very rare to have an electronic ballast/driver/power supply that results in such high neutral currents.

You seem to be stuck with this ballast idea. My point was not in reference to ballasts.