Sizing Generator with UPS 1

[sepanolo]

I have heard that when a stand-by generator is connected to a UPS, the generator should be rated at least 3 times the KVA rating of the UPS.

Does anyone know where this recommendation comes from?

I have a project with a 200 KVA UPS system which would require a minimum of 600KW worth of stand-by generator. It seems excessive.

Any info would be helpful.
thanks

 

[ScottyUK]

The generator will have to supply what is normally an awkward load, being laden with harmonics and at a poor power factor. This type of load heats up the generator, and makes life hard for the AVR. Oversizing the generator eases these problems.

The generator must also supply the UPS load, plus UPS losses, plus UPS charging current, which together add significantly to the load on the generator.

Some UPS's with active front ends claim to be able to wrk with generators oversized by only 20% or so - APC's Silcon range being one that springs to mind.

 

[busbar]

This problem has been around for decades, and can only be resolved with careful procurement verbiage. If not, the two vendors end up being a finger-pointing exercise, with the end user left with a marginal system. Oversimplified—the generator folks want to have their equipment serving a totally resistive load, and the UPS guys expect their system to be served by a zero-impedance source.

A solution may be specifying a single vendor to supply both with some sort of interoperability performance commitment.

 

[rbulsara]

This is an issue blown out of proportion, mainly by generator suppliers who do not see the complete picture of a project.

Consider the following:

The basic (misplaced) reason given is UPS loads produce harmonics on the line side and therefore the generator has to be oversized to mitigate potential excessive heat generated by the harmonics. While this sounds OK at first glance the practical reality is nowhere near it.

1. Almost all good UPS systems will have harmonic filters built-in and the input harmonics are normally limited to 7%, and definitely not more than 10%.
2. When a sizeable UPS is to be supported, it is not the 100% load on the generator. Because almost all critical loads (assuming computer type loads) requiring UPS needs cooling. The supporting loads including the cooling load is normally equal in kW to that of the UPS loads. This fact alone makes the 10% THD on line side a non-issue as the UPS load is now less than 50% of the generator capacity. Any motor loads also dampen the effect of harmonics at the generator terminals.
3. Battery charging is usually 10% of the UPS’s rated current, losses may be 10% so allowing for about 25% capacity for such items is more than adequate.
4. Now you factor in the normal cushion or margin thrown in for sizing the generator, say another 25%, the significance of the UPS harmonic diminishes even further.
5. Also if you are too technical, you can look into the probability of mitigating effect of harmonics, by using a proper pitch of the generator windings. For example 2/3-pitch winding is not affected by triple harmonics and so on. This is a slipper route as the winding pitch has other ramifications if you are paralleling the generators or paralleling with the utility company supply.
6. A fact of real life, I have yet to see a UPS system loaded more than 50% of its rated capacity. The harmonics, if any, are function of the actual load.

Considering all the above, you will have a generator sized more than twice the rating of an UPS system at any rate and if that is the case, I would not worry too much.

 

[busbar]

Vendors’ sometimes creative interpretations of an operational system, versus the end user’s and design engineer’s version of the same do not always conveniently mesh. The matter has to be addressed well before equipment is delivered to the site—in contract/procurement verbiage. Power ratio rules-of-thumb serve is a reality check, but may not assure a happy client.

If the combination of equipment doesn’t fill the needs of the client over an adequate range of operational conditions, you better believe that the respective vendors will, if not first trying to drop the problem like a hot potato, do their best to delay resolution of the problem. An assertion of a ‘defective’ excitation system versus ‘defective’ line rectification often get dragged out to result in a bailing-wire compromise. Each camp vehemently insists they have done nothing to upset project scheduling, particularly if the client does not have their own significant experience in the interaction of these two “imperfect” systems.

 

[alehman]

I agree with most of the above. Simply oversizing is not a good fix. I know of many systems where UPS loads account for 60 to 70% of the genset nameplate without problem. Yet other systems where the UPS is only 30% have voltage regulator problems.

In addition to the problems arleady mentioned is UPS input power factor. Many UPS's have very large capacitors on the input for harmonic filtering. If these capacitors remain connected at light load, the effect will be a leading power factor. That causes difficulty for genset voltage regulators and can result in voltage instability or shutdown of the genset. Some UPS manufacturers have recognized this problem and have means to automatically (or manually) disconnect some of the capacitance during light load conditions. When load is light, harmonic currents are low so harmonic filtering is not as important. As the load increases, the filters can be switched on and the reactive power is abosrbed by the rectifiers so the genset doesn't see leading power factor. I bring this up based on experience commissioning a few systems.

I'm not sure about rbulsara's item 5. 2/3 pitch minimizes triplen harmonic voltage distortion of the generator alone, but also minimizes impedance to triplen harmonics. This can have the effect of increasing triplen harmonic current. The best solution depends on the goal.

 

[Gideonv]

Some Merlin Gerin UPS's allows you to limit the power the UPS draws when it is running on generator power and to have a gradual startup. This would however mean that you are drawing more/most of your power from the batteries - and would mean a limited standby time.

http://www.mgeups.com/products/pdt230/largeups/galaxy/galapw.htm

We like this feature as it will allow us (at least in theory) to put a UPS in a site with limited generating power.

 

[alehman]

Power limit on generator is commonly set to provide adequate power to support the load while minimizing battery recharging power. Many larger UPS's have timed rectifier ramp-up to improve generator performance. If there are multiple UPS's ramp-up can be staggered as well.

 

[e5fornow]

I am currently involved in trying to fix a UPS with an undersized generator. So I have an idea where you are coming from. The reason the generator needs to be oversized is due to the nature of the UPS. When normal power fails the UPS will block load the gen-set. Most UPS have a max. of a 1 HZ/sec slew rate. Which means if your gen-set's freq. sags too much or too quickly your UPS will go back on battery.
The problem is not with the harmonics generated by the UPS most modern UPS have input isolation transformers. I know that the Toshiba and Active Power units approach unity power factor on the inputs.
The problem occurs when the engine rpm drops due to the block loading of the UPS. UPS are only about 80% efficient so therfore the generator is going to see about 250kVA from the UPS load, then you have the battery chargers and cooling loads.
To get the most out of your emergency generator, make sure you get an engine with an electronic governor and a voltage regulator with 2 volts/HZ regulation.
Also make sure your ATS supplies a signal to the UPS when it is on emergency power, your UPS can lower its recharge rate to reduce the load seen by the gen-set. Also have your AC units turn off until the UPS is charged.
I hope this helps. I do believe you could get away with a 350kw genset as long as the only load was the UPS.

 

[alehman]

"...Which means if your gen-set's freq. sags too much or too quickly your UPS will go back on battery."

I wouldn't think the rectifier would care about slew rate. The slew rate limit (df/dt) specifies how quickly the inverter output frequency can track the incomming line frequency. If the slew rate is exceeded most UPS's will drop off of the line tracking mode and go revert to their internal clock. At that point the static bypass function can no longer be used in a closed transition mode, so most UPS's will show an alarm condition "bypass not available". It should not load the battery at that point however.

In the case of voltage, the inverter doesn't normally track the input but is manually adjustable. Again when the difference to the line voltage becomes to large, it will go to "bypass not available".

The system should only load the battery if the input frequency or voltage is out of spec for the recitifier

It would be interesting if the original poster could respond with status.

 

[DaveMarz]

First thing you need to do is make sure that your UPS is designed to communicate with a genset. Once the UPS knows that it is on generator there are usually several parameters that you can set. IE the soft start time, frequency following envelope, and battery recharging. The other thing to watch out for is the harmonic filter. The data sheets you will see show the THD with the optional harmonic filter. I sell UPS units for a living and have lost bids to unscrupulous competitors who don’t include the filters in order to keep their price down.

Finally there are manufacturers who can provide you with genset UPS packages. You might want to look at this option.

 

[jbartos]

Suggestion: I agree with the previous posting that power quality "subtleties" can become costly to add when overlooked in the beginning of the project engineering or design.