Non-Linear Load

[GoldDigger]

Do you consider the neutral in modular furniture in an office to be a data center?
I do not have a link at hand, but there are documented cases if smoke and more from the furniture.
Interestingly enough, I do not recall the same situation overheating the branch circuit neutral.
Possibly a wire size issue too?

 

[jumper]

My final answer:

Maybe..possibly...it depends...sometimes...sorta...it could be...........

I hope that kid gets detention.

 

[Carultch]

With nonlinear loads combining on the neutral at a level somewhere between the arithmetic sum and a typical vector sum, how do you determine whether the neutral ampacity is exceeded before you put it in?

I think you'd need to connect it to an oscilloscope that displays its Fourier transform. And add up the currents at of each harmonic that is a multiple of 3 of the fundamental 60 Hz frequency.

That is the main reason the NEC has this concern about nonlinear loads. Because 3rd and multiples of 3rd harmonics do not cancel among balanced phases. They accumulate. And 3rd harmonics are very common in a lot of loads.

 

[don_resqcapt19]

Do you consider the neutral in modular furniture in an office to be a data center?
I do not have a link at hand, but there are documented cases if smoke and more from the furniture.
Interestingly enough, I do not recall the same situation overheating the branch circuit neutral.
Possibly a wire size issue too?

Show me it was caused by harmonics...I have never seen one that was documented showing harmonics as the cause.

 

[GoldDigger]

It was determined that the load was almost entirely desktop computers and that none of the individual lines were overloaded.
So I would say that it was clearly established that the phase conductors were not overloaded but the neutral was.
There may be other possible causes for that (such as a combination of leading PF on one phase and lagging on another) but harmonics were the only credible suggestion.
I believe that a reconstruction was done to establish liability, but I never saw those results published.

 

[Haji]

With nonlinear loads combining on the neutral at a level somewhere between the arithmetic sum and a typical vector sum, how do you determine whether the neutral ampacity is exceeded before you put it in?

An arithmetic sum would suffice for practical purposes.

 

[ActionDave]

It was determined that the load was almost entirely desktop computers and that none of the individual lines were overloaded.
So I would say that it was clearly established that the phase conductors were not overloaded but the neutral was.
There may be other possible causes for that (such as a combination of leading PF on one phase and lagging on another) ......

Or a lousy connection on a lousy piece of mass produced office furniture.

I'm with Don on this one.

 

[Smart $]

An arithmetic sum would suffice for practical purposes.

That would be the extreme on the large end of the range and would mean the circuit current is 100% triplen (i.e. third harmonics equal to or greater than fundamental). Odds of that being the case is---as a guess---quite, quite unlikely. Even "beefed up" panels only go 2x neutral.

 

[Sahib]

That would be the extreme on the large end of the range and would mean the circuit current is 100% triplen (i.e. third harmonics equal to or greater than fundamental). Odds of that being the case is---as a guess---quite, quite unlikely. Even "beefed up" panels only go 2x neutral.

Not the triplen but take the non-linear load current say it is I1 in phase A and I2 and I3 are linear load currents in phase B and C. If the arithmetic sum 'I1+I2+I3' [with I1/(I1+I2+I3)<50%] is equal to the ampacity of the neutral conductor, no problem. But if I1/(I1+I2+I3)>50%, the neutral conductor is to be de-rated, even if the sum 'I1+I2+I3' is equal to the ampacity of the neutral conductor.

That is my take.:happyyes:

 

[kwired]

Can anyone show me a documented case of neutral conductor overheating from non-linear loads in someplace other than a data center?

This is a case where almost all of the published information on the problems that may be caused by non-linear loads come from the people that have something to sell you to solve the problem.

:thumbsup::thumbsup:

Even in a data center is it much of a problem on multiwire branch circuits or primarily a problem with feeders and service conductors?

What I call a data center from my work experiences is nothing compared to what some of you guys run into out there.

 

[Smart $]

Not the triplen but take the non-linear load current say it is I1 in phase A and I2 and I3 are linear load currents in phase B and C. If the arithmetic sum 'I1+I2+I3' [with I1/(I1+I2+I3)<50%] is equal to the ampacity of the neutral conductor, no problem. But if I1/(I1+I2+I3)>50%, the neutral conductor is to be de-rated, even if the sum 'I1+I2+I3' is equal to the ampacity of the neutral conductor.

That is my take.:happyyes:

That's essentially the same as what I said earlier... only under NEC the proper method is VA_nonlinear / VA_total.

However, when phase B & C are receptacle loads, there's no guarantee they will not be nonlinear.

Haji's post was not clear in what he wants to arithmetically sum....

 

[Tony S]

Normally I_n = √((I_a²+I_b²+I_c²)-((I_a*I_b)+(I_a*I_c)+(I_b*I_c))) takes care of the natural triplen harmonic.

Introduce non linear loads then the other neutral harmonics come in to play.

It’s not just the neutral wiring you need to worry about, you’re likely to have one very unhappy transformer. Type K transformers are becoming more common in the UK due to failures caused by neutral harmonics.

 

[Sahib]

I committed blunder in my post#29. Apologies to forum members.