Sizing a Neutral for Parallel UPSs

[jerebear]

We are installing 2 - 3 phase UPS modules in parallel to a common SBM. The manufacturer of the UPS is requiring a neutral being pulled in each of 4 parallel feeders from the UPS to the SBM. This neutral will not be bonded and will carry no load as all loads are 3 phase 3 wire. This neutral is only being used as a reference between the 2 UPS modules for synchronization. The manufacturer is requiring a full size neutral in each of the feeders and when asked why they stated that they only require it because they believe the code requires it. The code does allow for reducing the size of the neutral based on the unbalanced load but in reality this neutral will never carry any load. How do you go about sizing this if indeed it is to be reduced?

 

[DRWeig]

1) Read up on article 215 in the NEC.
2) Catch this article: EC Mag
3) Consider the cost of arguing, convincing the UPS supplier, and convincing the AHJ versus the savings by downsizing that neutral.

Wait for other responses, I may be all wet.

PS: What is SBM? Google says it's a Single Black Male, but I bet he wouldn't want parallel feeders hooked to him.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[DRWeig]

PPS - if you can justify calling that neutral a "control signal," you can downsize it a lot.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[iop995]

Are your sure that UPSs are symetric load? Input rectifier of UPS is a very distorsion load even with a good PFC. This type of load will impose same neutral current. Much worse in case of rectifier diode demage.

 

[sibeen]

iop995, read the OP. The load is a three wire system.

Saying that, what effect does an input rectifier have on an output neutral? Ie, none.
Also,as this seems to be a new system I'd imagine that it would have an active front end and so the distorion will be quite low, and the power factor quite high. Not that either of these have any bearing on the current in an output neutral, especially in a three wire system

 

[jerebear]

Here is an explanation I received from one of the engineers that works for the manufacturer.

"These are circulating currents between the UPSs. They are not current being delivered to the load.
When paralleling UPSs, there are multiple inverters that are constantly adjusting their outputs to match the system output. The voltages won’t be exactly the same which induces current in the neutral conductor.
There are other scenarios that also induce current flow through the neutral. For example, if there is a load step, you may have a momentary imbalance on how the phases are loaded. In this case you would have a spike in the current through the neutral conductor."

 

[sibeen]

jerebear, and that is a fair point. If you ever monitor the current out of a single UPS, that is operating in a parallel configuration, the current will be jumping all over the place. The output load of the sytem as a whole will be steady, but individual UPS currents, not so much.

There's a fair bit of smoothing, or interpolation, done by the UPS output current meters

 

[DanEE]

Quote: "These are circulating currents between the UPSs. They are not current being delivered to the load.
When paralleling UPSs, there are multiple inverters that are constantly adjusting their outputs to match the system output. The voltages won’t be exactly the same which induces current in the neutral conductor.
There are other scenarios that also induce current flow through the neutral. For example, if there is a load step, you may have a momentary imbalance on how the phases are loaded. In this case you would have a spike in the current through the neutral conductor."

This makes sense and also makes a point I was wondering about the design of the UPS in question... The UPS in question must be similar in design to for example, some of Alpha Technologies medium sized UPS systems in that (for purposes of discussion, assume this is a 208 3 Phase system) the UPS generates 3 phase power by the use of 120VAC modules running off a master clock with, of course, a bank of 120VAC modules generating Phase A at the 0 phase angle, the second bank of 120VAC modules timed to generate Phase B at 120 degrees and the third bank timed at the 240 degree point, for Phase C.. So you have a Y source (and from what you stated), driving a delta load.

Paralleling two UPSs of this design is no different that if the manufacturer supplied a single system with expansion cabinets for additional generating capability.. i.e. parallel rows of 120VAC modules for each of the 3 phases, all running off a common master clock and in a Y configuration.. and harmonics, imbalance loads can be flowing in the interconnecting neutral between the 2 systems up to the limits of the system, (if in worst case load). Down sizing this neutral, based on what the customer is doing today is risky, as who knows what load will be on the system tomorrow, in terms of imbalance and harmonic content. Will the manufacturer agree in writing to support the warranty on this system if not installed per their installation specs? Will AHJ approve if the installation is not in compliance with the mfg installation specifications?

While maybe not relevant to load configuration in this discussion, but given todays extensive use of switching power supplies in load equipment (e.g. server farms) and with the subsequent harmonics they generate back to the source, Alpha requires neutrals in 2 and 3 phase systems to the load to be doubled in size in relation to the phase conductors. Reason? With significant harmonic content encountered in 2 and 3 phase loads, the harmonic frequency content of current in the neutral can be in-phase, and additive, opposed to fundamental frequency current which cancels as we expect. Even if the customer load does not have this problem at installation time, the system must be installed to handle what they put on it in the future.

 

[sibeen]

DanEE, it used to be quite common for the neutral to be oversized. Certainly I saw plenty of sites, ten to twenty years ago, where the neutral current exceeded the phase current. All to do with triplen (zero sequence) harmonics, predominately the third.

Today it is not an issue. Power supplies have improved in leaps and bounds, and the harmonic content drawn is way down, and the power factor is way up. I wouldn't ever bother specifying that a neutral be oversized in any new installation.

 

[jerebear]

Based on the latest information from the manufacturer we have gone ahead and pulled full size neutrals in the parallel feeders. Also as recommended by the manufacturer we have not bonded the neutral. Thanks everyone for your input. Very interesting reading.