UPS Nuisantly Transfers to Bypass 5

[nightfox1925]

I am presently working on a problem with one of the UPS systems in the plant. There are two Single UPS systems (480-208/120) feeding two UPS buses (A and B) separately. The two separate bus circuits feeds a couple of 208VAC-24Vdc rectifiers which in turn individually feed as redundant power supplies to downstream PLCs, SIS, etc.

Of the 24Vdc rectifiers, the 24Vdc, 400A rectifier is the largest. The UPS systems always shifts to bypass everytime the UPS output MCCB is closed or this Rectifier is energized. The UPS alarm shows "Output Overload". We suspect that it was the downstream in-rush which is causing the UPS to nuisantly transfer. An idea by one of the contractor was to put a reactor on the input of the rectifier. A 5% impedance reactor was connected in series but still the UPS keep to transfering to bypass.

It is observed that the UPS output transformer secondary is wye and the downstream rectifier isolation transformer primary is delta...is it possible that the problem is also caused by phase mismatch between the UPS and downstream charger?

is there a possibility that the UPS settings may be too sensitive?

I am gathering all the data so I can post them here soon. But any help or guidance will be very much appreciated. Thank you.

 

[sibeen]

Going back to the series resistors. Did they call these "buffer" resistors before? Is the concept actually putting the resistors in series while energizing the transformer so as to create a voltage drop on the line, consequently reducing the voltage input at the transformer primary and avoid saturating the transformer primary to decrease the starting flux generated in the core based from formula:

*caveat* Lots of hand waving explanations and not much maths.

Nightfox, have a look at the small single line I have included. The resistors are placed in-line to the transformer and are shorted out by the contactor K1 after energisation of the transformer. What you need to do is size the resistors to such a size that the current into the transformer is going to be smaller than the current that will force the UPS to bypass. If, for example, we make the three resistors one ohm each, then the current into the transformer will be limited to approximately 100 amps.

By doing this the length of the inrush will be extended and may last as long as a second or so before the transformer reaches its ‘steady state’ magnetisation level. For a 20 kVA transformer I’d guess that this would be about 5 amps. So, the wattage of the resistor needs to be sized to be able to handle a quite considerable amount of power for a short duration and then a reasonably lower level of power (25 watts?).

The problem with this is that the rectifier must not be allowed to turn on before the contactor has closed, otherwise you will then be trying to draw maximum load through the resistors (10000 watts), this will produce smoke in large quantities. So the timing of the contactor closure must be long enough to ensure that the transformer has been substantially magnetised and short enough to ensure that it closes before the rectifier has a chance to start.


To all, can anybody explain how the "current limiting" function of the UPS works...for my benefit of understanding.

There is a difference between "current limiting" and overload operation in a UPS. The current limiting function is normally only used when the Bypass of the UPS is not available. In this case, if there is an overcurrent on the output of the UPS, the inverter of the UPS will then ramp down its voltage so that a maximum current is provided at the output. If there was a short on the output, for instance, the output voltage may go as low as a few volts to limit the output current.

Overload capability of 150% at 10 seconds, is the 10 seconds count inherent to the UPS even if the UPS is set at 100% overload?

I’m not sure what you’re asking here. Whilst I’ve never seen a Mitsubishi UPS, this is my take on it. If the output load is >110% and <125% and the bypass supply is available, then a timer will start in the UPS and transfer the load to bypass 10 minutes later.

If the load is between 125% and 150% the same timer will transfer the UPS to bypass in 10 seconds. With a load >150% the UPS will transfer to bypass immediately.

 

[nightfox1925]

We finally was able to complete the site measurements at site using a high speed analyzer. We made tests with the following scenarios:

1. Rectifier is energized with DC output breaker on

2. Rectifier is energized with DC output breaker OFF (DC
loads and batteries disconnected).

3. Rectifier is energized with DC output breaker reclosed
after 20 minutes (to partially discharge the batteries)

Among the 3 scenarios, item 3 gave us the worse current up to 430A peak to peak. The rectifer AC input is rated for a full load of 48A at 208VAC. This results to approximately 8.9X or say 9X (9 P.U.) as inrush.

The upstream UPS in this location was having an output rated for 75kVA at 208VAC. The overload capability for 150% is 112.5kVA for 10 seconds. The inrush equates to 154.9kVA pk-pk which is 206% (2.06P.U.)! and this is not even including the other 208VAC continuous loads from then other panelboards which amounts to 56.2kVA. Adding them up to the inrush kVA will result to 154.9+56.2 = 211.1kVA, a whooping 281% which totally exceed even the 200% rating at 1 sec. This is why the UPS instantaneously transfer itself to bypass.

In sizing the appropriate mitigation, am I going to use the in-rush peak-to-peak or the equivalent in-rush RMS?

Thanks

 

[waross]

I am not sure but may I make a suggestion.
The UPS rating is probably RMS. The peak will probably not be a sign wave. How about converting the UPS's RMS rating to peak and then comparing to the peak current. Use peak to zero rather than peak to peak. If this is a transformer inrush the current will be asymmetrical. Peak to peak will ignore the DC component which should not be ignored.

At this point, now that you have hard numbers, can you contact the manufacturer to determine if this is acceptable operation for this UPS? Ask in writing and for a written response.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[nightfox1925]

Hi Waross,

The official measurement results will be provided by the Third Party testing company tommorow. I will definitely send it to Mitsubishi including a copy of the wave form. I will scan the wave form and post it here as well.

For the mean time, do you have any canadian based transformer, current limiting reactor/resistor manufacturer for me to contact?

Thank you for your patience and assistance.

 

[nightfox1925]

Hello everybody,

We have received the measurement results and waveform for the in-rush measurements for the downtream rectifier. This rectifier is connected to a 100kVA, 208VAC, 277.6A, 0.8P.F. UPS inverter at its output. The UPS transfers to bypass everytime the large downstream rectifier is energized.

The other UPS loads in addition to this large recitifier is a 2.521kVA rectifier and a 208VAC UPS panelboard with connected load of 56.206kVA.

The downstream large rectifier is having an input of 208VAC, 60Hz, 3Ph, 60A and an output of 24Vdc, 500A and connected to a 12cells, 755Ah storage batteries.

The In-rush currents along with their respective scenarios are as follows:

1. The Recitifer is energized with its 24vdc Loads +
batteries connected. Batteries are fully charged

Time: 11:36:21
Phase A: +420Arms, +594Apk
Phase B: -560Arms, -792Apk
Phase C: +340Arms, +481Apk

2. The Recitifer is energized with its 24vdc Loads +
batteries DISABLED (Converter DC output MCCB OPEN)

Time: 11:41:42
Phase A: -460Arms, -650Apk
Phase B: -620Arms, -877Apk
Phase C: +500Arms, +707Apk

3. The Recitifer is energized with its 24vdc Loads +
batteries CONNECTED (Batteries partially discharged for
10 minutes).

Time: 11:45:18
Phase A: +80Arms, +113Apk
Phase B: +160Arms, +226Apk
Phase C: -140Arms, -198Apk

How can we possibly distinguish and determine how much component of the in-rush is real and how much is reaactive? I understand that if the in-rush has a significant load (real) component, a resistor mitigation is effective, if the magnitizing in-rush is significant, then a reactor will be effective. I appreciate any further comments.

 

[sibeen]

nightfox, I'm not sure why you think that the resistor wont work if the current is reactive.

Ohms law still applies. If you have 208 volts and 2 ohms of resistance you going to have a maximum current of 104 amps.

 

[nightfox1925]

I guess I misinterpreted something here. My bad. You're right Sibeen, the total equivalent impedance is the vector sum of the resistance and reactance, so it does not matter whether its purely resistance or reactance.

Looking into waross statement:

"BUT we still should know the cause of the inrush. If it is a reflected load from the DC bus, and is real current, you want resistors limiting it. If it is magnetizing inrush it may be highly reactive and respond to reactors."

Sibeen, do you have any comment on this so I could be further enlightened? Is this statement has something to do with the current displacement?

 

[nightfox1925]

Waross, if we were to size a series reactor to limit the in-rush, which measured in-rush values will be used, the RMS values or the Peak Values?

 

[waross]

I think that I should defer to one of our protection experts.
I think that there are others here with experience with current limiting reactors. I have a good idea, but I don't want to lead you astray.
For a symmetrical inrush I would use the RMS, Or 0.707 of the peak value.
For an asymmetrical inrush I would spend several hours that I can't spare just now doing some study and a lot of cut and try calculations on the DC offset issues. I would be looking at the phase angle of the AC component and trying to combine the resistive voltage drops of both the AC and DC components and the the reactive voltage drop of the AC component.
I don't have the time just now but I am sure that we have some posters here who work with current limiting reactors and will be able to supply a better answer quicker than I on the sizing of reactors on asynchronous surges.


Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[nightfox1925]

Thank you very much for all the guidance. I am actually, looking into several options to mitigate the issue. Although I fully support the idea that placing a proper start up procedure for the downstream rectifier and appropriate switching (since redudancy of 24Vdc supplies are in place)will tend to make the present set-up operational.

However, this is not the kind of perception that the operations are having and further convincing has to be done. I am also considering other options like placing current limiting reactors and a low in-rush isolation transformers. I will coordinate with the manufacturers of these equipment with regards to properly sizing them with the available In-rush data and continuous load data and as well as the operating limits of both UPS and rectifier. All advantages and disadvantages will be part of the study so the client will understand the operating limitations and possible hickups in choosing each option.

Waross and Sibeen, thanks for all the technical comments, it helped me understand the situation. Waross, thanks for reminding me about the UPS manufacturer with regards of getting their side with this actual in-rush data. I may post further separately on the deal of selecting and sizing reactors and hope I will also get opinions on that subject matter from the associated experts of this forum.

 

[nightfox1925]

By the way, it is stated here that there is no issue of a UPS output transformer whose secondary is wye and feeding a rectifier whose input transformer is having a Delta primary winding.

What if the UPS output transformer whose secondary is wye is feeding a 3Ph, 4 wire panelboard and then the delta connected rectifier is fed from this panelboard via 3 wire and the panelboard is feeding other line-to-neutral loads?

Is this going to increase the inrush drawn by the rectifier input transformer?

 

[davidbeach]

Nope, not a problem.

 

[nightfox1925]

Thanks david. There are rectifier or UPS manufacturer who strongly believe that phase mismatch is a cause for the high inrush. However, they can't present anything when I asked them to justify this in terms of physics and calculations. Thye just based it from their "experiences". Unfortunately I don't want to place this phase matching isolation transformers by trial and error.

I am curious though what made almost all of them think that it what they think it is...

 

[norman63]

What is the frequency tolerance setting on the UPS?

 

[alehman]

nightfox,
I don't understand this "phase mismatch". Are they saying that having a wye connected transformer serving a 3-wire load somehow creates high inrush?

 

[sibeen]

By the way, it is stated here that there is no issue of a UPS output transformer whose secondary is wye and feeding a rectifier whose input transformer is having a Delta primary winding.

Surely all, or at least the vast, vast majority of three phase, on-line, UPSs have an output transformer which has a wye secondary?

There are thousands of computer centres around the world where a 'standard' UPS with a wye output transformer is feeding an isolation trandsformer at a PDU, or similar, which has a delta primary.

In a lot of these cases the UPS will transfer to bypass when energising the transformer. This is becauses of the inrush current to the transformer and has nothing to do with some percieved phase miss-match.

 

[nightfox1925]

norman63, to answer you question, the UPS free running output frequency (asynchronous or on battery) is 60Hz, +/-0.01%

 

[norman63]

Check the frequency pot setting. I've had problems in the past with machines when trying to sync to generator power, batteries and/or outside power sources/draws. You should be able to set the tolerance to plus or minus 5%. 1% is a very tight setting considering your set up and load. Try backing it off 1% at a time until the units stop(s) transfering into bypass. Are both of the UPS units Mitsubishi? Do they have internal batteries, external batteries or both? What size are the batteries? Are the rectifiers your feeding part of a D/C plant (batteries included or are you relying on the UPS for battery power?) Are all UPS, Rectifier and/or D/C plants new installations and are the batteries new? What is the float voltage set at for the UPS units? Are both UPS untis 100KVA? I will dig into this a little deeper at my end. I have excellent tech support available and they should be able to nail the problem down quickly. Try what I suggested first and send me the info on the machines involved.

 

[Handlin]

As a application engineer for a UPS manufacturer (not Mitsubishi), it appearts to me the issue is UPS sizing. There is nothing in the thread that indicated there is a problem with the Mitsubishi UPS. The UPS is a 100KVA box @ .8 PF - meaning the 100% full load is 80 KW. At roughly 105% load (84 KW) the unit will transfer to bypass if the bypass is avaiable. The best solution is to determine the maximun load (including inrush) and size the UPS according. Industry standard is for the design load to be approximately 80% of the full load rating of the UPS - so for the 100KVA UPS about 80 KVA/64KW.

 

[raithrovers1]

I agree with Handlin. The UPS is not faulty at all and is doing exactly what it should be doing. The problem is in the sizing of the UPS.
That does not cure yor problem though. May I suggest that you look at installing a soft start unit between the output of the UPS and the input to the rectifier. This will control the inrush to the system by gradually increasing the output voltage available to the rectifier input therefore limiting the inrush current to the transformer. This will prevent the UPS tripping to bypass, and also is a far better and safer way to limit inrush that using a high wattage resistor and contactor circuit.
Vist this site for an idea:

http://www.softstartuk.com/

I am a power electronics engineer working for the following company:

http://www.powerups.co.uk