Feeding UPS from generator

[Volta]

So it seems that there is a limit to the portion of a generator that can be allotted to UPS loads.

For a few small low-end UPS units on a standby generator, we are finding suggestions to plan for a maximum of 15-20% of a generator's capacity to be UPS loads. That is because a simple UPS is likely to cycle on and off with voltage inconsistencies as the various supplies see the restoration of (generator) power, load the gen, and pull down the voltage.

Does that sound right?

A better double-inversion UPS might be less sensitive, but we are still finding percentage limits suggested.

If our client has approximately 6kW of loads through about half a dozen existing UPS units, plus say a limit of .5kW for lighting, how small of a generator can we install to provide a reliable system?

Are we really talking about a 30/27 (fed with natural gas) generator for these cheap supplies to run well?

 

[weressl]

So it seems that there is a limit to the portion of a generator that can be allotted to UPS loads.

For a few small low-end UPS units on a standby generator, we are finding suggestions to plan for a maximum of 15-20% of a generator's capacity to be UPS loads. That is because a simple UPS is likely to cycle on and off with voltage inconsistencies as the various supplies see the restoration of (generator) power, load the gen, and pull down the voltage.

Does that sound right?

A better double-inversion UPS might be less sensitive, but we are still finding percentage limits suggested.

If our client has approximately 6kW of loads through about half a dozen existing UPS units, plus say a limit of .5kW for lighting, how small of a generator can we install to provide a reliable system?

Are we really talking about a 30/27 (fed with natural gas) generator for these cheap supplies to run well?

You may want to search teh forum for previous posts on the subject, but the gist of it is this:

UPS's are designed to operate supply a fairly narrowly defined voltage/frequency window to the protected equipment. When the incoming voltage is outside of those parameters, the UPS will turn the incoming power off and start supplying power from the batteries. The larger and more sophisticated the UPS the more flexibility it will offer for adjustability in these parameters, as to what is the acceptable window.

Generators come in all sizes, obviously the smaller units are employing simpler voltage and frequency regulation, so the UPS's may cycle on/off, eventually exhausting the battery capacity. The other issue is the nature of thye UPS front end, what tytpe load does it represents to the generator. Most of them are linear loads with fairly high harmonic content, so your voltage/frequency regulator has to be built for tolerating that, otherwise it could just go wild all over and haunting for the stability condition that never comes. (You could also have a significantly high(12-14x) inrush that will also cause haunting.)

 

[Besoeker]

For a few small low-end UPS units on a standby generator, we are finding suggestions to plan for a maximum of 15-20% of a generator's capacity to be UPS loads

Why would you want to feed a UPS from a standby generator?

 

[weressl]

Why would you want to feed a UPS from a standby generator?

Routinely done. UPS receives normal supply and when the normal supply is interrupted the ATS starts the standby generator and after the voltage stabilized, trasnfers the UPS normal input source to the standby generator.

Two benefits:

• Reduced battery bank size,
• UPS generates heat. If it is operated in a small building and as the power is lost the HVAC is not available, the UPS overehats and shuts down even though there is still sufficient battery power to continue supplying the load. So the generator is sized to support the UPS, the HVAC and whatever else it does need to supply.

 

[Volta]

So is it reasonable to conceive of some type of filtering system, without batteries, on the output of the generator to improve the quality enough that it would be allow the cheap existing UPS units to accept the generator power?

As if the ATS panel were fed from a high quality UPS, so the various point-of-use supplies would be happy, but the filtering might be 1 or 2 grand instead of 4-5 grand for a good large UPS.

 

[weressl]

So is it reasonable to conceive of some type of filtering system, without batteries, on the output of the generator to improve the quality enough that it would be allow the cheap existing UPS units to accept the generator power?

As if the ATS panel were fed from a high quality UPS, so the various point-of-use supplies would be happy, but the filtering might be 1 or 2 grand instead of 4-5 grand for a good large UPS.

May or may not work, you need to put numbers to it and calculate the dynamic results, but you would need to have detailed technical data from each and not easy to come by. It's even possible that a single large buffering transformer would do the job, but hard to tell. The larger your generator is in relationship to the individual units, the better chance you got.

 

[jdsmith]

A better double-inversion UPS might be less sensitive, but we are still finding percentage limits suggested.

If our client has approximately 6kW of loads through about half a dozen existing UPS units, plus say a limit of .5kW for lighting, how small of a generator can we install to provide a reliable system?

Could you replace the numerous UPS units with one good one purchased from a manufacturer that could provide application engineering to answer the generator sizing question? Are the existing UPS's old enough that you can justify replacing them? Are they the small cheap ones with with sealed lead acid batteries that need to be replaced every 3-4 years, regardless of amount of use?

Unfortunately most of the UPS's on the market today come with poor documentation, poorly trained salesmen how can't answer any technical questions, many set points in the units are not very programmable, and the documentation is fairly poor so we can't really understand how they work.

You have one of the best UPS companies in the country in your back yard - Ametek Solidstate Controls. If you have the budget and floor space for a UPS replacement it might be worth some investigation. I'm not sure what a 30 kVA genset costs so I'm not sure how much savings is available. The other consideration is battery maintenance - vented lead acid or ni-cad cells will last 15-30 years compared to 3-4 years for sealed lead acid, so there is a payback available there if your client is interested.

 

[Volta]

May or may not work, you need to put numbers to it and calculate the dynamic results, but you would need to have detailed technical data from each and not easy to come by. It's even possible that a single large buffering transformer would do the job, but hard to tell. The larger your generator is in relationship to the individual units, the better chance you got.

That's the direction we were heading with it, but that gets to be a relatively big generator.

Could you replace the numerous UPS units with one good one purchased from a manufacturer that could provide application engineering to answer the generator sizing question? Are the existing UPS's old enough that you can justify replacing them? Are they the small cheap ones with with sealed lead acid batteries that need to be replaced every 3-4 years, regardless of amount of use?

Unfortunately most of the UPS's on the market today come with poor documentation, poorly trained salesmen how can't answer any technical questions, many set points in the units are not very programmable, and the documentation is fairly poor so we can't really understand how they work.

You have one of the best UPS companies in the country in your back yard - Ametek Solidstate Controls. If you have the budget and floor space for a UPS replacement it might be worth some investigation. I'm not sure what a 30 kVA genset costs so I'm not sure how much savings is available. The other consideration is battery maintenance - vented lead acid or ni-cad cells will last 15-30 years compared to 3-4 years for sealed lead acid, so there is a payback available there if your client is interested.

Thanks, we'll check out Ametek too. I do prefer the idea of one new decent unit, even if it is $5k, that is a savings over a double size generator, I think.

Yes, I think the existing units seem fairly cheap, with SLABs, and on average are probably 4 years old I'd guess. While the batteries aren't that costly to replace, if we can give them a 20 year lifespan, that's sure an added benefit.

 

[ceb58]

The generators we have have been sized 2x larger than the UPS's. The most of our UPS's are 18Kw but only have around 5-7Kw load on them. All of ours are duel conversion units and they do not like bad Hz. They have a tolerance of +/- 2Hz so when running on gen. the UPS will switch over to batt. due to the start up of the A/C unit then they go back to gen. power. I have been told by the UPS engineers that you can figure your load on the UPS and add 15% for the charger. That should give you a real close draw on the unit.

 

[tkb]

When you put a UPS on a generator you will get better performance if you add some resistive load in addition to the UPS load.

 

[Volta]

The generators we have have been sized 2x larger than the UPS's. The most of our UPS's are 18Kw but only have around 5-7Kw load on them. All of ours are duel conversion units and they do not like bad Hz. They have a tolerance of +/- 2Hz so when running on gen. the UPS will switch over to batt. due to the start up of the A/C unit then they go back to gen. power. I have been told by the UPS engineers that you can figure your load on the UPS and add 15% for the charger. That should give you a real close draw on the unit.

I figured that there should be some rule-of-thumb for that, 15% sounds reasonable to me when unsure of the makes/models.

I don't know what the current load exactly is (about 6 server/desktops, CRTs, phone system, fiber transceiver), the various UPSs nameplates total 5.7k, I think, so that is what we're working with. No other loads contemplated yet. We'll consider room for expansion, minimal lighting, and the inevitable coffee maker after we have the main plan, and see what that does to the proposal. The customer was pretty resistant to 'extras', so I'll want a firm foundation before we mention again hvac, sump, or similar. They expect that if the network stays up, the various agents and other sites can function from different locations.

It is not a 'server room' per se, rather it is a large basement, so I don't feel that cooling is an issue here.

When you put a UPS on a generator you will get better performance if you add some resistive load in addition to the UPS load.

I was wondering about that. We are looking at a decent load capacity available with these percentages, will additional resistive load actively help to correct the generated waveform somehow? That would help with the basic lighting/coffeepot suggestion, if so.

How about inductive but linear loads? Any benefit there?

 

[dkidd]

There is a good article in ECM

http://ecmweb.com/news/electric_ensuring_generator_ups/index.html

 

[Volta]

There is a good article in ECM

http://ecmweb.com/news/electric_ensuring_generator_ups/index.html

Thanks, some good food for thought there, there is a lot for me to learn on this project!

 

[masonc]

I have just spent a lot of time on this subject while taking a generator course at Generac's training facility and discussing this with the Generac engineers. Generac will normally suggest no more than 20% of the generator's load be a UPS's rated capacity. Bear in mind, the UPS will likely not be fully loaded, so it may be that the load is only 50% so your generator is sized 10x the actual load. Oversizing is no guarantee of success as there are a number of factors involved.
Gaseous generators are particularly prone and having the UPS as a single load makes the situation worse. Generators that use SCRs to switch excitation power will be more prone to this problem. PGM based voltage regulation is also better as it forms an isolated voltage sensing source.

The reason for the difficulty in powering the UPS load is to do with the way a UPS controls power. The UPS uses SCRs to chop the incoming waveform, switching the wave off when it has enough power from that cycle. The generator's voltage regulator see that as reflected harmonic and this can destabilize the voltage regulator and send the generator into a shutdown mode.
Additionally, a UPS will likely have a bunch of power conditioning capacitors on it's input to remove the incoming noise. When the UPS disconnects because of voltage variations, which will happen with a generator, it leaves the line conditioning in place. The generator will see this as a completely reactive load with a leading power factor. This again can destabilize the AVR and send the generator into a tail spin.

Generators that use SCRs to switch the exciter power depend on zero crossing sensing to accurately control the voltage to the excitation coils. SCRs in the UPS switch around that point as well and will distort the waveform at zero crossing. This again will disrupt the operation of the AVR in the generator and cause a control loop incompatibility.

Some UPS may use different technologies that allow a higher load factor. However, it is unlikely the UPS manufacturer can guarantee the performance of the generator/UPS combination so you could be very embarrassed on startup. Even identical setups at different locations may have different results. I would get the UPS and generator set manufacturers involved in the design.

Chris Mason
Comet Systems