UPS testing at full load

[jlchaves]

Hi all!

We're purchasing a new 16kVA UPS and want to test it before signing off any acceptance documents.

We usually simulate full load by using a dummy load consisting of electrodes immersed in a tank of water. Salt is added to the water until the UPS is at full load.

I'd like your comments:

1. Does anyone here use the same method to test your UPSs?

2. Up to how high a load should I test? I'm contemplating testing up to 110% load for up to 3 hours.

3. What other stuff should I test? I will probably hook up a digital oscilloscope to capture the waveform.



Thanks,

Jan

 

[davidbeach]

I once ran the testing of a 300kW UPS using 200 - 1500W personal space heaters as the test load. Allowed the room HVAC system to be tested at the same time.

 

[itsmoked]

200!!!!!!!!!!!! I thought my 100 light bulbs were absurd... How did you plug them all in???

What was that, $8,000 worth of heaters?


jlchaves; That load of yours is purely resistive you might want to add some sort of inductive loading. Drag in a 3HP motor and toss it on too. Maybe a wheel around air compressor. Let it run thru an open valve.

 

[davidbeach]

It was a fully fit up computer room, lots of receptacles under the floor, didn't use anywhere near all of the receptacles. I only suggested the heaters, the contractor provided and I have no idea what the cost. I think they came in a pallet load from Costco. Could use fewer, larger, heaters but then you have to worry about cooking a sprinkler head.

Most UPS systems designed for computer system use don't like motor loads much, I'd be careful about using motors as a test load.

 

[MikeHalloran]

If you're including a transfer switch and a generator in the system, verify that the UPS will in fact pick up and sustain the load while accepting power from a generator.

You're thinking, 'Duh'. I did, too. But my two small UPS units would not allow their respective computers to start while the AC power was coming from a generator. Both computers started and ran just fine from the generator without the UPS.

I was too busy patching the roof to figure out why the UPSes did not care for generator power. It was clearly not undervoltage.



Mike Halloran
Pembroke Pines, FL, USA

 

[Warpspeed]

There can be all manner of stability problems when attempting to drive a UPS from a generator. The problem is that the generator will have some sort of feedback loop to maintain a reasonably constant output voltage with changing load. And the UPS is also attempting to maintain a constant output voltage. Both these feedback systems will have time constants, and can sometimes interact in unfortunate ways.

Rather a lot depends on if the UPS is the only generator load, or just a small part of a much larger fairly constant resistive load.

The worst type of UPS to drive with any sort of generator, is the ferro-resonant type. As input voltage falls, current rises so it looks like a negative resistance load to the poor generator. Also, small generator speed and frequency changes can cause fairly drastic voltage and current changes at the load, because the ferro is "tuned".

All this can create some horrible stability problems that can be difficult if not impossible to cure. It may all work perfectly, or it could turn out to be quite an "interesting" day.

 

[jlchaves]

Thanks for the ideas, guys. I rely on building gensets to provide auxilliary power in the event of a blackout.

Here's another challenge: I've been wanting to install a small genset for years, but we are located high in a commercial building. Short of punching intake and exhaust ports in the walls, I don't have a clear-cut way of installing a genset. Fuel cells cost too much, too.

Jan

 

[MikeHalloran]

Thanks, Warpspeed.

Jan, disguising the gas ports is probably achievable, but concealing the presence of a generator from your landlord and/or insurance company carries some obvious risk. Safely storing fuel near the generator, and replenishing it when the elevators don't work, are bigger challenges.

Scattered through this blog:

http://nola.us/frame.html

you will find references to buying, storing, guarding, hauling, transferring, filtering, and reeking of, Diesel fuel.



Mike Halloran
Pembroke Pines, FL, USA

 

[jlchaves]

Thanks, Mike. Guess I'm in a real fix. Probably will resort to colocation OR stocking LOTS of batteries.

The worst part is the building engineers normally cut ALL power to perform maintenance during certain holidays here in the Philippines. I have to shut everything down for 3 days. Even the chilled water supply is shut off, so my cooling system dies, too.

Bummer

Jan

 

[itsmoked]

jlchaves; What are you doing there that you'd like to do on weekends too? Maybe we can offer some other solutions.

 

[jlchaves]

Hi there itsmoked. I'm actually working off my PDA. Weekends are my only free time to write in the forums.

Anyway, we're pushing through with the UPS purchase soon. We're just finalizing the designs.

If the budget fits, we're going for a 20kVA unit that runs on 3-phase power.

Jan

 

[MikeHalloran]

<brain fart>

You could lower a heavy weight, braking it with a generator, and hoisting it up again when power is restored. Just like winding a cuckoo clock. Depending on the building, you might be able to drop the weight on the outside, increasing your energy storage capacity.

A gargoyle that very slowly changed floors when the building was unoccupied might go unnoticed for a long time...

</fart>





Mike Halloran
Pembroke Pines, FL, USA

 

[RonShap]

In NYC, and many other cities with large buildings, the generators live on the roof with a day tank.
I also commission UPS's with heaters in the room (to test the HVAC) or load banks outside (on ground/roof). With many new computer power supplies compliant with the IEC regulations, power factor is relatively high and harmonics is relatively low when the computers are connected.
You can also rent a resistive/reactive load bank for hours of testing fun.

 

[RudyL54]

What an ingenious way to test it with water and salt as dummy load.

I haven't used water with salt.

How do you know how much salt you need to add to measure the full load and what should be the increment?

 

[itsmoked]

You just look at the amp meter and dribble in more salt.

Salt water rheostats work well (if not great). They will get hot. You can end up with lots of steam. They are sort of self regulating since if things get too wild they self increase in resistance.

Do be careful you don't melt your plastic container..

 

[mpparent]

Maybe a dumb question, but what about a load-bank?

Mike

 

[ruggedscot]

When we ran up our new build and did some test proving on out brand spanking new data hall we used 608 3kW heaters split up over the three phases fed to the data hall. Total load ran to about 1.8MW. Cool ? Nah it was hot ! this tested the power distribution, the UPS units and also the heat management in the hall ensuring that it kept within the specified limits. We had some fun swinging the loads around and trying as best as we could to make it fall over, but it didnt....

Now its a functioning data hall and we cant do what we did back then but it sure as heck kept its end of the deal.

Rugged

 

[itsmoked]

608... further amazement..

 

[rcw retired EE]

If you are using a salt bath, make sure that the corrosive steam and slat water stay away from any of the UPS equipment or anything else that you don't want to rust quickly.

You can rent water-cooled load banks.

I have worked in many high-rise buildings that had a common diesel generator either on the roof or in the basement for several tenants. I wondered how the building manager decided who was overloading the unit when problems occurred?

 

[jlchaves]

Haha, nice brain fart, Mike. Perhaps a windmill might work, too.

About our dummy load, we just get the largest nonmetallic tank we can find. I think the last time we used a 50 gallon plastic barrel, the kind used for fruit juice concentrates. It did get pretty warm, but not too hot to melt. The water was hot but not boiling. Wooden spacers kept the electrodes apart.

We met with the UPS vendor yesterday. We finalized the design at 20kVA, N+1. Which means three 10kVA power modules. Input is 3ph.

I'll probably need three 50-gallon barrels and a lot of cable for this. Breakers everywhere too.

Jan