Installing a UPS on a UPS 2

[dug1dug]

I'm maintaining a data center here in Texas, we have a Liebert UPS (4 750KVA modules per UPS with redundancy). In talking with building engineers, Leibert Techs, other electricians, i have learned you are not supposed to install a UPS on a UPS. My question is why. The only answer i get is "you just don't". The reason I'm asking is there are a number of cabinets, same make, with three UPS's in each one that give us quite a bit of grief. The owners of the cabinets insist on having the UPS's. They are not very reliable, and they drag our power factor way down, less than 70% in cases where they are the only equipment on a Power Distribution Unit. The owners want to blame our power supply, but they get the same protected power as anyone else on the floor yet they are the only one with problems. So if you shouldn't put a UPS on a UPS any body know why?

 

[ykee]

Most UPS's are installed in Data centers for 2N or at least N+1. I'm not sure how you would hook up a UPS to provide back up to a UPS. If you have 2N or N+1 set-up, you can have UPS failure and still be up.

Have you put your UPS's in series--so that if one fails, nothing gets through (do you have a single-line?)?

I have not heard of a UPS that causes a bad power factor.

 

[itsmoked]

You don't put them in series. If the power fails the first one then is faced with charging the second and running the load. Also some may put out a fine-for-running-equipment waveform but strange to the second one causing it to take over.

Also as ykee sez you can end up with one taking both out.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[dug1dug]

Yes we do have 2N+1, but we have some computer equipment out on the floor with their own ups inside the cabinet, even tho we supply protected power to them. These are small units about 3000VA that power only the individual cabinet. Its these individual cabinets we're having problems with.

 

[alehman]

I've also seen rack UPS's in the data center. If there is a properly designed large UPS upstream, I think they can only serve to reduce the overall reliability. I don't know why the rack UPS would have a problem though. You might ask that manufacturer.

 

[SteveFehr]

The output filters on one UPS can have unwanted interactions with the input filters on the other. It can increase harmonics and circulating current and reduce component life. I say "can" because, as I understand it, it's very fuzzy, difficult to predict, and doesn't seem to cause any really obvious problems that one can point to, but still can cause damage to both units.

Bear in mind, that those rack-mount UPS will add additional charge current after an outage, so if you do lose the lieberts and they come back up, those little UPS are going to go into recharge simultaneously, and you'll see a big current spike.

 

[Senselessticker]

The extra "in-line" UPS only reduces the overall reliability of the complete system. It introduces something "extra" that can fail... Most users/owners of the specific equipment have trouble understanding...

 

[sibeen]

There is no hard and fast rule about feeding one UPS from another and I have designed systems where the output of one UPS will feed the bypass/reserve input of a second UPS.

In the case that is mentioned by du1dug, I'd be getting rid of the small UPSs. As others have mentioned the reliability of the system will be less than the large UPS system alone.

As for the small UPS manufacturer claiming that the input power to their small units is causing problems, it sounds to me as if they are trying to shift responsibility. A 3kVA unit running off a 3MVA large UPS should have absolutely no problems. Have someone throw a scope on the PDU unit to have a look at the waveform and check the voltage THD to make sure that there is no weird reasonance issues. Present the figures to the manufacturer and tell them to get their **** together.

 

[Senselessticker]

Send the individual rack owner a bill for the additional cost they are inducing on the system by having an extra (and futile) UPS to serve their equipment.... OR... tell them that if they want to be responsible for their own power redundancy... you'll be happy to provide them a few circuits from a panel that is not UPS / Gen. backed...

 

[LionelHutz]

^^ I agree, feed them raw power if they insist on their own UPS. Only give them the 2 options. If they want protected power then they have to get rid of their UPS.

 

[itsmoked]

You're not thinking like a business man Mr Hutz...

If they say that and the user pulls his UPSs and then you fail to provide continuous power,(and it happens), then you get all the blame. If the customer has his own UPSs and anything at all goes wrong you get to point at his UPSs.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ruggedscot]

This is one that is cropping up in data centres

Integral UPS units in a rack can cause issues with the supply and you have to be really careful as you could cause problems with other kit on site.

1. Id let them know that your UPS network is proven and there has never been an outage on your site. Explain if you do have an outage how does that affect the infra structure on site like cooling and comms to the racks ? No good having a unit protected if it cant interact.

2. Is it for controlled shut down ? Some servers do have this fitted so that if there is a power outage then the server can detect the utility failure and then shut down safely.

3. emergency power off buttons - these dont link with the kit should a power down be initiated then the rack will ramin live until the batteries run down. Not good if you have the hall EPO system that doesnt kill all the power.

4. Introducing batteries into a hall - electrical supplies are protected and fused. Put a large battery into the hall and you have a power source that can not be isolated.

5. Did I say batteries ? If you have a leak of the electrolite that wrecks the underfloor infrastructure.

6. Did I say batteries ? If you have a fire in the data hall you have a risk present from a battery.

7. Id ask him for a full risk assesment for having batteries in his equipment and then ask him for reassurance that his kit will not have an effect on your kit. Id also see about setting up a standard specification for equipment within the hall. Only kit that is to your specification is allowed in the hall - a bar being placed on non conforming equipment. That way you can keep control.

If your UPS and system is properyl designed then there is no need for a UPS to be within the hall.

Rugged

 

[electriken]

If this is in an "Information Technology Equipment Room" per NEC Article 645...
...645.11(2) limits the size of a UPS in the room to 750VA

 

[davidbeach]

limits the size of a UPS in the room to 750VA

No it doesn't. 750VA is the maximum size that doesn't have to meet the disconnect requirements of 645.10. I've done redundant 300kVa UPS systems and larger in the 645 room, no problems.

 

[electriken]

davidbeach is indeed right - I stand corrected.
By the time the customer pays for the reconfiguration of the EPO System, additional rack space, and a cooling surcharge to accommodate there unnecessary heat load they could possibly have a change of heart.

 

[ruggedscot]

How about this as a thought - forget a UPS install a static switcher and provide an alternative supply path to your unit. Thus the unit is fed from two independant supplies one being set a primary source and the other as back up should the prime supply fail the unit switches over to secondary without break. Load protected and no ugly ups in your room.

Good thing is these units are passive and dont require batteries. 2nd supply has to be seperate from the primary source say another UPS else where?