A customer wants to install a 1000kw standby generator to serve several HVAC motors and three 150kw UPS units for his computer room. I was concerned that the combination of voltage drop and harmonics during start-up, would cause the UPS's to reject the source and drain its batteries. Has anyone experienced any incompatabilities between generators and UPS's? Thanks!
Generator serving UPS - problems?
- Thread starter cosmos
- Start date
- Status
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
Generic answer- check with the genset and UPS manufactureres 
More specific answer- Do not try to run a UPS on a mechanically-governed genset, although I'm not sure how common mech gov's are for newer installations. The governor can't keep up with minute load changes from the UPS and it can start whacking the throttle around. It's interesting to watch. From a distance.
I assume these are 3-ph input UPSs.
With the sizing described and the 3-ph motors, I think you should be fine. It will help, of course, if all the loads don't try to climb on the genset at once. I recall that some UPSs have soft-start inputs. Do the customer's units have this, or is an upgrade available?
More than UPS, are there any larger ferro-resonant transformers or single phase UPSs out there? Those can play hob with generators.
More specific answer- Do not try to run a UPS on a mechanically-governed genset, although I'm not sure how common mech gov's are for newer installations. The governor can't keep up with minute load changes from the UPS and it can start whacking the throttle around. It's interesting to watch. From a distance.
I assume these are 3-ph input UPSs.
With the sizing described and the 3-ph motors, I think you should be fine. It will help, of course, if all the loads don't try to climb on the genset at once. I recall that some UPSs have soft-start inputs. Do the customer's units have this, or is an upgrade available?
More than UPS, are there any larger ferro-resonant transformers or single phase UPSs out there? Those can play hob with generators.
dbuckley
Senior Member
- Location
- Canterbury, New Zealand
If the A/C units are packaged units (common for computer rooms) then they often have a menu setting for a start-up delay, so you set the first one at 30 seconds, and then the rest them 15 seconds apart. Then when the ATS brings the genset on line the loads dont "try to climb on the genset at once" - like that phrase.
A double conversion UPS will try very hard to keep the output syncronised in phase with the input, and most big UPSs have menu settings (often service personnel only) to control how "fussy" the UPS is in this respect. The reason the UPS tries so hard is because of bypass. If the UPS self-checks fail it'll go straight into bypass. More importantly, if there is a significant step increase in output current or an overload (eg need to clear a downstream fault) then the UPS will go into bypass, as UPSs have very poor fault current generation capability, usually only about 150% for one cycle, which isn't enough to rapidly clear a fault. So they go to bypass, and let the upstream source (normally utility power) supply the fault current, and then the UPS goes back online, all over in a couple of seconds. They cant do this if the output and input is out of sync, or the input is missing, in which case the UPS usually collapses and dies requiring manual restart.
Computers are not terribly fussy in terms of input power, and can tolerate an unsynchronized static bypass switchover, unlike motors, which really don't like that kind of thing. Thus you can allow the UPS to go quite wide in terms of frequency and rate of change of frequency.
Assuming theres a bit of slack on the loads of the UPS, the genset is only going to be loaded a bit above 50%, so an easy ride, but not too underloaded.
A double conversion UPS will try very hard to keep the output syncronised in phase with the input, and most big UPSs have menu settings (often service personnel only) to control how "fussy" the UPS is in this respect. The reason the UPS tries so hard is because of bypass. If the UPS self-checks fail it'll go straight into bypass. More importantly, if there is a significant step increase in output current or an overload (eg need to clear a downstream fault) then the UPS will go into bypass, as UPSs have very poor fault current generation capability, usually only about 150% for one cycle, which isn't enough to rapidly clear a fault. So they go to bypass, and let the upstream source (normally utility power) supply the fault current, and then the UPS goes back online, all over in a couple of seconds. They cant do this if the output and input is out of sync, or the input is missing, in which case the UPS usually collapses and dies requiring manual restart.
Computers are not terribly fussy in terms of input power, and can tolerate an unsynchronized static bypass switchover, unlike motors, which really don't like that kind of thing. Thus you can allow the UPS to go quite wide in terms of frequency and rate of change of frequency.
Assuming theres a bit of slack on the loads of the UPS, the genset is only going to be loaded a bit above 50%, so an easy ride, but not too underloaded.
brian john
Senior Member
- Location
- Leesburg, VA
In the early days of UPS?s this was an issue, with improperly sized gensets, and as noted mechanical governors. For the most part this has been resolved. But depending on the type of UPS (and options ordered), there can be input filters, this can lead to some issues during very fast transfers when both sources are available, normally when returning to utility power (depending on ATS style).
There are some options that are utilized with motor loads and UPS operations, as noted above, step starting equipment and current limiting recharge current of the batteries while on generator.
There are some options that are utilized with motor loads and UPS operations, as noted above, step starting equipment and current limiting recharge current of the batteries while on generator.
360Youth
Senior Member
- Location
- Newport, NC
zbang said:Generic answer- check with the genset and UPS manufactureres
More specific answer- Do not try to run a UPS on a mechanically-governed genset, although I'm not sure how common mech gov's are for newer installations. The governor can't keep up with minute load changes from the UPS and it can start whacking the throttle around. It's interesting to watch. From a distance.
Even the smaller gensets we are putting in are electronicly governed. There are more and more of them. We have never had to size for that kind of UPS system, but we have had to size for hydraulic and similar elevators that are very sensitive to voltage drop. With proper sizing and ATS technology, the right genset is out there.
27hillcrest
Senior Member
If all the equipment to be installed is going to be new then you shouldn't have any issues. If the UPS's are going to be 4x2 units then keep in mind the inrush current for the bypass XFMR if a maintenance bypass panel is used. There should be curve study done on the system to make sure the generator and the main breaker will accept the inrush from the bypass XFMR. Also when you transfer from generator power to street power. I have witnessed this failure and doesn't look good when you are trying to certify the system to the owner.
catchtwentytwo
Senior Member
I've been taught that inductive loads (like the motors on AC's) should be put on the bus right after any life safety loads to help stabilize the generator. Besides, in a "race" between overheating a computer room and draining a large UPS battery, overheating usually wins. Be careful of not delaying the AC restarts too much.
Most large UPS systems can receive a control signal that the building is on generator which can be use to ramp or limit the battery recharge current. That recharging current can be substantial and bog down a generator.
Most large UPS systems can receive a control signal that the building is on generator which can be use to ramp or limit the battery recharge current. That recharging current can be substantial and bog down a generator.
dbuckley
Senior Member
- Location
- Canterbury, New Zealand
The process coolers are ok for a minute or so to get back on line at full capacity, but what some folks have yet to appreciate is that specific cooling for blade servers and other high density equipment has gotta be on the UPS; a rack with 20KW of load in it cannot be deprived of cooling, of it will fail in seconds, and probably quite impressively. You cant wait the ten seconds for the genset to spin up and the ACB to close; its already too late...catchtwentytwo said:
27hillcrest
Senior Member
dbuckley said:
I have been working in an 11,000 square foot data center for the past few months updating all the cooling equipment and adding 225kva UPS. One night the new CRAC units shut off do to the drain line freezing up outside because the heat tape wasn't turned on. The room went form 71 to 95 five in about 25 mins if things went much longer things would have been real bad. The room has only about 40 10U Dell blade servers. So things do heat up very fast.
robbietan
Senior Member
- Location
- Antipolo City
the best deal seems to be check how much charging current the UPS is rated then see if the genset can cope up with this. the charging current plus the inrush current of the motors may be too much for the genset to handle
dbuckley
Senior Member
- Location
- Canterbury, New Zealand
25 minutes... yeah, thats a long time, but a few minutes is ok. Sure, you get a bump on the graph, but not enough to be a problem. In the OPs case he's got about 450 KVA of UPS, and probably about a dozen 50KW chillers, and at 30+15 secs the lot (which is more than necessary, redundancy and all that.) would be back online within five minutes. You can make the the times closer, and the bigger the site then the closer they can get, even down to a second or two, as the genset is correspondingly more able to deal with inrush of the chillers. Of course, if it's a water cooling with central chiller system, then it's all a bit different.27hillcrest said:
450KVA is typically a corporate style data centre, and these don't get pushed excessively hard; it's too small to be a colo full of crazy kit.
But, bottom line, cooling is the most important service for a computer room. Lose the power, you lose the power, and you moan, and look forward to when it's back, all the while telling the customer what a bunch of idiots the sparkies are to lose power like that. On the other hand, lose the cooling, and then you (err .. perhaps I should say I, and people like me) have to turn stuff off pronto, before it frazzles. Then we need to explain to customers why we turned their boxes off, and that conversation is much harder, I assure you
catchtwentytwo
Senior Member
I experienced a short cooling failure in a 26,000 sq/ft computer room that had approximately 500 KW of UPS load none of which were blades or any other especially dense watts per square foot hardware. It is astounding how fast things heat up.
ghostbuster
Senior Member
- Location
- Great White North
Another area of concern,could be the UPS harmonic filtering and motor rotational load inertia(acts like a generator) interaction on the same bus system during transfer.On some designs, we have spec'd in midposition time delay transfer switches to allow for the normal electrical decay times to occur.Just a thought.
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
FWIW, I've seen people squalk long and loud about connecting HVAC loads to the 'technical' power (UPS), and there's merit to the argument that the other loads may polute the clean power (also, to the argument that you need a double-size UPS). OTOH, I can easily see putting the circulating fans on UPS, but not chillers/pumps/compressors/anything with a VFD.
Since you -need- temp monitoring anyway in case an HVAC unit or two dies, the fact that the HVAC stops for 30 seconds isn't of too much concern to me,
especially if room air is still moving. Longer than that, and we'll talk.
Since you -need- temp monitoring anyway in case an HVAC unit or two dies, the fact that the HVAC stops for 30 seconds isn't of too much concern to me,
especially if room air is still moving. Longer than that, and we'll talk
.dbuckley
Senior Member
- Location
- Canterbury, New Zealand
Oddly enough, to a UPS a VFD looks just like a computer, as under the covers, a VFD and a SMPS are the same technology. Fear not the VFDs on UPSs.zbang said:
I did work at one site where there were three UPSs and no genny; two of the UPSs were for standard Liebert chillers. The logic weas they could shut the site down in 20 minutes, thats all the contract required, and the UPSs would hold the computer room complete with chillers for that time.
It was a smallish site, 120KVA main UPS for computer loads, and (I think) 80KVAs for a pair of Lieberts each. These Lieberts had DOL starts for the motors, no VFDs, so a fairly nasty load for a semiconductor UPS, as they don't like big step changes in output loads, but it all seemed to work, it operated that way for seven years I know of.
hillbilly1
Senior Member
- Location
- North Georgia mountains
- Occupation
- Owner/electrical contractor
I have had issue's with smaller gensets (10-20 kw) and small portable ups units (350 VA and typical) used in small home offices. The ups will cycle constantly while on generator. I don't know if its frequency or fluctuations in the voltage, but they don't like it. Some ups units can be reprogrammed to not be as sensitive to the fluctuations.
The main issue with UPS is that the load that they are serving. It is widely kwnown that often UPS supply a data center with a load of Blade type server, the power factor could be leading unless the hamonic issue is solved.
If the UPS is equipt with Active Hamonics conditioner (AHC) then should be an isue for this.
If the UPS is equipt with Active Hamonics conditioner (AHC) then should be an isue for this.
cosmos said:
Years ago, I saw a backup generator feeding a big bank of UPSs get into that surge and stall cycle. It was quite unnerving to watch as the lights and AC in the computer room would dim and brighten with the generator, the UPSs meeping plaintively as they switched from full battery to generator power every few seconds.
We shut down all the computers as fast as we safely could, and then the UPSs. Then the generator started running normally, to our stupificaiton. When we started turning the UPSs back on, the generator started surging again. Nobody around could explain why it had happened, and the electricians called in to investigate were scratching their heads too.
Someone finally heard about the feedback thing, and the quick and effective solution was to double the size of the genny, and put more conventional loads on it. So the executive decision to put all the break room fridges, vending and coffee machines on the emergency generator was made, and the next time the power went out, the employees were a lot less disgruntled.
Check with the generator manufacturer to ensure it is rated to power computers and data centers, and check with the UPS manufacturer to see what they say about operating their UPSs on backup generators.
One very interesting backup power system I've heard about, and would like to see sometime. It has eliminated the batteries from the UPS. Outside you have a box on a pad. Inside the box is an appropriately sized diesel engine, clutch to a shaft that is normally spun by a line powered motor. On the other end of the shaft is the generator that poweres the computers. Shaft has a biggish flywheel, and a compressed air powered air turbine that has enough compressed air tankage to spin the thing at full load while cranking up the diesel engine. Air compressor to recharge the air bank.
No more hugely expensive and hazardous battery banks that have to be replaced every couple of years and are loaded with hazardous materieals. Just a bank of high pressure air flasks that have to get inspected and pressure tested every five years.
If you need more capacity, you add more air flasks. I like the concept a lot.
We shut down all the computers as fast as we safely could, and then the UPSs. Then the generator started running normally, to our stupificaiton. When we started turning the UPSs back on, the generator started surging again. Nobody around could explain why it had happened, and the electricians called in to investigate were scratching their heads too.
Someone finally heard about the feedback thing, and the quick and effective solution was to double the size of the genny, and put more conventional loads on it. So the executive decision to put all the break room fridges, vending and coffee machines on the emergency generator was made, and the next time the power went out, the employees were a lot less disgruntled.
Check with the generator manufacturer to ensure it is rated to power computers and data centers, and check with the UPS manufacturer to see what they say about operating their UPSs on backup generators.
One very interesting backup power system I've heard about, and would like to see sometime. It has eliminated the batteries from the UPS. Outside you have a box on a pad. Inside the box is an appropriately sized diesel engine, clutch to a shaft that is normally spun by a line powered motor. On the other end of the shaft is the generator that poweres the computers. Shaft has a biggish flywheel, and a compressed air powered air turbine that has enough compressed air tankage to spin the thing at full load while cranking up the diesel engine. Air compressor to recharge the air bank.
No more hugely expensive and hazardous battery banks that have to be replaced every couple of years and are loaded with hazardous materieals. Just a bank of high pressure air flasks that have to get inspected and pressure tested every five years.
If you need more capacity, you add more air flasks. I like the concept a lot.
dbuckley
Senior Member
- Location
- Canterbury, New Zealand
Not seen one with air, but the Pillar rotary UPS systems is somewhat similar, see http://www,piller.com
The engine and motor-generator set is on the left, and a flywheel energy storage device and switchgear on the right. When the power fails the flywheel storage unit keeps the generator rotating whilst the diesel starts, whcih then takes up the load.
One really neat feature of these units is you can upsize the diesel engine, and the motor part of the motor generator can be a generator when the engine is the prime mover, so you can have additional generation for your aircon.
The engine and motor-generator set is on the left, and a flywheel energy storage device and switchgear on the right. When the power fails the flywheel storage unit keeps the generator rotating whilst the diesel starts, whcih then takes up the load.
One really neat feature of these units is you can upsize the diesel engine, and the motor part of the motor generator can be a generator when the engine is the prime mover, so you can have additional generation for your aircon.
- Status