Load Imbalance

[davidbeach]

In a project I am working on, I have a 208Y/120V unit substation supplying a load that is essentially all 120V, 20A, receptacle circuits - many panels, each with 30 or so 20/1 breakers. The design loading is reasonably well balanced from phase to phase, the bulk of the load being 180VA receptacles.

Does anybody have any empirical data as to how much imbalance I might see in actual operation? The worst possible case would be, with the building fully loaded, somebody went through and turned off and the phase 'B' and phase 'C' breakers, giving 100% load on 'A' and nothing on the other two. But that wouldn't be very likely. On the other hand, there could be, by happenstance mostly, a noticeable unbalance. How unbalanced could one reasonably expect to see?

Thanks.

 

[alehman]

My experience is overall general building systems without any large single-phase loads are balanced within 20 or 30%. Sometimes single phase equipment such as air conditioning units, heaters and large copiers cause imbalance at individual panels.

 

[dpc]

My experience is consistent with alehman's. Of course, if the concern is neutral current, you have to consider third harmonic current which can easily exceed fundamental neutral current in 208/120V systems.

 

[busbar]

The description “load being 180VA receptacles” is fairly meaningless. Would you elaborate on what's plugged into the receptacles?

 

[davidbeach]

busbar, the building will be a university School of Engineering, loads will be offices, instructional spaces, teaching labs. So the loads will run the gamut but will primarily consist of switch-mode power supplies. All of the larger loads are fed at 480V off of other unit substations. I'm not sure that the type of load really matters though, it's more an issue of the degree of randomness in what is actually running at the worst imbalance.

The concern is not neutral loading, but rather negative and zero sequence currents in the three phase conductors on the primary of the 12,470V circuit feeding the grounded wye-grounded wye transformer of the unit substation.

 

[alehman]

"wye-grounded wye"? Is your 480V system is to be grounded? If so you need "wye-grounded wye-grounded".

You may wish to consider using delta primary to stop propogation of triplen harmonics.

What specifically is the concern for negative and zero sequence currents with the transformer? Transformers can operate with any amount of unbalance.

 

[jbartos]

Comment: If the 12.47kV transformer side is grounded and low voltage side is also grounded, the grounds may experience elevated potential during lightning if the 12.47kV line is overhead.
It is better to have the 12.47kV ungrounded (delta) to avoid the zero sequence current propagation and elevation of grounds during lightning on 12.47kV side.

 

[davidbeach]

alehman, it is grounded wye - grounded wye. I don't care about negative and zero sequence currents in the transformer in particular, but want to use them for detecting faults, faults which might be less than full load currents. Not a big deal if the fault detector function can't be set less than full load, but it should be possible.

jbartos, 12.47kV line is underground, and a decent ground grid is being installed.

The transformers are power cast with fan cooling to 150% of nameplate base rating. I have found it very difficult to do fully selective coordination on delta - grounded wye transformers and make full use of the 150% of full load. With a delta - wye (grounded or not) the damage curve is at 58% of the current of a wye - wye (grounded or not at either side). The local utility is not much worried about the propagation of triplen harmonics as all of their service transformers are grounded wye - grounded wye anyhow.

 

[busbar]

One description of the load may be ‘institutional’. With a grounded·wye grounded·wye transformer, secondary-winding [switchmode} triplen currents would be reflected into the primary neutral. I am not sure what the concern is as far as transformer load or thermal losses, with or without neutral current.

It could be that IEEE Std C57.96-1999 Guide for Loading Dry-Type Distribution and Power Transformers contains applicable information. [There is a table of contents for the 1989 edition at standards.ieee.org/reading/ieee/std_public/description/dtransformers/C57.96-1989_desc.html ]

 

[jghrist]

davidbeach,

From your responses, I gather that you intend to use zero- and negative-sequence relaying to gain sensitivity for Ø-Ø and Ø-grd faults and need to know the expected phase imbalance to determine how sensitive you can make the settings. For this purpose, the third-harmonic question is probably irrelevant because most microprocessor relays respond only to the fundamental frequency. You might want to confirm this with the relay manufacturer that you are using.

You could use the experiences of ahleman and dpc (20-30% imbalance). You're situation may have more imbalance because there may be times when a teaching lab or two on one phase are not in use while other spaces are fully used. You could monitor the load on a similar facility to get a better estimate.

 

[davidbeach]

jghrist, thank you. After ahleman & dpc's early responses I was beginning to wonder if I was getting the point of the question across to anyone. Yes, I am looking for sensitive fault detection of phase-to-phase and phase-to-ground (I need to learn how to get the special characters to show up here) faults. The relays in the scheme are a mix of SEL-351 and SEL-501-2. The 501s are on the outgoing feeders with the 351s on the mains, ties, and generator breakers.

I am looking to implement a fast bus trip scheme, sort of a poor man's bus differential, where outgoing breakers (which can include the mains and the ties depending on operational configuration, loads, and local generation) which see a fault block the fast trip of the next upstream breaker. The goal is to detect and clear any bus faults within 6-7 cycles while maintaining full selectivity.

Obviously any 3-phase fault will have to greater than maximum allowable load current as there is no way to distinguish a 3-phase fault from load. But any fault that does not involve all three phases can be distinguished from load through the zero- and negative-sequence currents.

If the maximum imbalance is in the range of 20-30%, and I've heard 30% elsewhere also, then I'll add some margin and set the fault detectors at 50-60% of the transformer full load rating.

 

[jghrist]

A lot of special characters can be created by holding the Alt key down and entering 3- or 4-digit numbers (128-255 and 0128-0255) with the numeric keypad. Alt 0216 gives Ø.

Any greek letter can be created by entering &[ig-nore]letter;[/ig-nore] (remove the "-" in ignore). Capitalize the written out letter to get a Greek capital letter. Omega for &[ignore]Omega;[/ignore] and omega for &[ignore]omega;[/ignore] for instance.

 

[BJC]

How big is the transformer?
The SEL-501-2 relays are protecting feeders to? 208/120 panels? Will there be intermediate distribution panels?

It is very difficult to coordinate the high side protection of a transformer. I don't like to do it but sometimes you have to use electronic fuses. There not cheap and you locked in to one supplier ( there may be more now ).

The 20- 30% imbalance is hard to design for. Some suggestions I have are.
1.) Spread the circuitry out, Put receptacles from at least 2 phases in each room or area. The cheap way to so it is to hang 10 receptacles on each circuit and locate then in one area. People get by with that for a while in spec built offices but i don't think it'll work in an engineering lab. Most of the office buildings designed like that get reworked sooner or later.

2.) Leave enough spares in each panel to relocate loads at a later date. Monitoring the operatiog conditions is the only real way to tell whats going on. The way things are used probably won't turn out to be what you or anyone else planned.

3.) To find out what is goin on for step 2 you have to monitor usage over time. Tha can mean temporary recording devices or better yet if the projcet has the money a smart metering system. Square D has power logic which is good but all the major equipment suppliers have systems that are just as good.

 

[davidbeach]

BJC, the SEL-501-2 is on a breaker on the primary of a 1000kVA 12470-208Y/120V transformer. Transformer feeds unit substation. Unit substation feeds five 1200A subdistribution boards (1 per floor). Each subdistribution board feeds an average of 12 or so branch panelboards. Each branch panelboard feeds receptacle loads, which by design are evenly distributed across all three phases.

But that's all in the theoritical world of design, where things are a balanced as possible. What I am interested in is how differently the real world behaves.

 

[alehman]

I like BJC's advice. It is very hard to predict actual loads. As equipment gets installed and moved around, loads and balance change.

Load imbalance will show up on negative sequence sensors, but zero sequence sensing should include neutral CT's to cancel the effect of normal neutral current.

SEL-501's are a little pricey. Have you looked at the zone selective interlocking capabilities in newer electronic trip units for low voltage breakers? I think ZSI can accomplish at least part of what you are wanting.

 

[davidbeach]

The SEL-501's are on 12.47kV breakers. I looked at trying to extend the low voltage ZSI up to the medium voltage level, but it would probably have taken a solid state interposing relay to take the very low level ZSI signal and made it usable as an input to one of the SEL relays.

 

[jbartos]

Suggestion: Normally, the larger power distribution with many diversified loads statistically balances to approximately 20-30% or better as posted in above postings. However, the power distribution implementation needs an experienced electrical contractor or electrical detailed design of circuits and outlets at downstream ends. The upstream and larger loads are mostly 3Ø (1Ø large loads should be avoided and not specified, if possible).
Even smaller projects, e.g. residential houses with 208V/120V 1Ø 3W, can have the load balanced, if 1Ø receptacles are balanced and larger loads, e.g. air conditioners, heaters are 208V instead of single phase 120V. Obviously, if there are all receptacles on one circuit, i.e. on one phase, in the house, the panel may become very unbalanced. However, this is the electrical contractor responsibility, if there are no electrical drawings.
Similar reasoning holds true for large projects. It may be too late to try to balanced loads, if the power distribution is improperly designed and installed by the electrical contractor. It may require an expensive re-work in some instances, (i.e. "pay now or pay later," which many of us very well know).
In old times, when there were fuses only, the rest of electrical power distribution was functional when one or two fuses were blown; approximately 33% unbalance for one fuse blown and 67% unbalance for two fuses blown.

 

[jbartos]

Suggestion to jghrist (Electrical) Jan 2, 2004 marked ///\\A lot of special characters can be created by holding the Alt key down and entering 3- or 4-digit numbers (128-255 and 0128-0255) with the numeric keypad. Alt 0216 gives Ø.
///The Numeric Lock light on the keyboard should be on\\Any greek letter can be created by entering &[ig-nore]letter;[/ig-nore] (remove the "-" in ignore). Capitalize the written out letter to get a Greek capital letter. Omega for O and omega for ? for instance.
///Use a dictionary for Greek letter spelling in the Greek alphabet.\\

 

[davidbeach]

Why do I have to keep repeating this? The design is balanced, but that balanced load is something on the order of a total of 4000 duplex receptacles. Could be well over a thousand with nothing plugged in at any given time. Much of the rest could be off if the room in question is not in use at the given time. I know how to circuit up a building for a load that calculates out as balanced, and as such it won't get far enough out of balance for any phase to be overloaded. That's all just basic design.

What I've been looking for is the difference between that balanced, calculated, design loading and what might happen, worst case, when the users of the building have some portion of the potential load on and the rest off. So far it looks like an unbalance of 30% is the highest likely to be seen, so I'll set the fault detectors above 50% to leave some margin.

 

[jbartos]

Suggestion to the previous posting: The original posting does not indicate how is the fault detector specified for the power distribution. Are you writing specifications (conceptual or engineering part of the project) or doing design (design phase of the project)?