UPS for IT + VFDs equipment simultaneously.

[namesdontmatter]

I have a data center that wants the UPS to power both IT equipment and HVAC equipment (VFDs/VFDs primarily). I have told them my concerns about having motors and sensitive IT equipment on the same UPS, but they have argued that the VRF systems have soft start, and won't impact the IT equipment. They have done this previously on another build, and I've seen the power quality reports from that build. Highest THD U is ~4%, 5th and 7th harmonics 2-3%. But that building is not fully loaded, so I can't say this won't be a problem in the future.

Does it matter if the HVAC system uses VSD/VFD or not? My intuition tells me any additional non-linear load will create power quality issues.

 

[Arester]

Harmonics will be contributed and will affect power quality. All depends on the quality of the VFD and the source/UPS impedance to load ratio. In data centre I would say it’s a bold approach, certainly cost effective … if things go well. Your intuition works like mine. Have a good weekend.

 

[winnie]

I don't have field experience with this, so only a question:

The IT equipment takes AC power, rectifies it, and then uses switching technology to develop the appropriate internal voltages for the various internal systems.

The VFD HVAC equipment takes AC power, rectifies it, and then uses switching technology to drive the various motors at appropriate speed.

In particular the VFD systems won't create the huge surges associated with across the line motor starting.

VFDs operate at much lower frequencies than the IT switching power supplies, and might inject some amount of switching noise into the electrical system.

But other than the different switching frequencies I'd expect VFD HVAC systems to 'look' very much like IT system loads.

What am I missing?

Thanks
Jon

 

[petersonra]

I don't see how the IT equipment is going to care about harmonics at all. The computer systems are well protected by the DC power supplies that power them.

 

[texie]

My question is why one would put the HVAC on a UPS.

 

[petersonra]

My question is why one would put the HVAC on a UPS.

Otherwise the computers will overheat if you have a power failure. It doesn't do any good to power up the computers if they're going to just shut down due to heat. On the other hand I really think that the HVAC stuff could be put on a generator rather than a UPS but that would seem to be a design decision and up to the engineer who made the decision..

 

[texie]

Otherwise the computers will overheat if you have a power failure. It doesn't do any good to power up the computers if they're going to just shut down due to heat. On the other hand I really think that the HVAC stuff could be put on a generator rather than a UPS but that would seem to be a design decision and up to the engineer who made the decision..

Of course they will need cooling. Just not from a UPS. If this is a small installation where maybe there is not a genset usually the UPS just powers the data equipment long enough for a graceful shutdown and the HVAC is not needed to run. On the other hand if this a larger install where the data equipment needs to run longer term there will always be genset and you connect the HVAC to the genset. I have never seen HVAC equipment connected to a UPS as it is an incredible waste of very expensive UPS power.

 

[namesdontmatter]

I don't see how the IT equipment is going to care about harmonics at all. The computer systems are well protected by the DC power supplies that power them.

Some computer equipment have voltage, frequency and THD requirements. IEEE 519 recommends general systems have no more than 5% THD U. IEC 61000-3-2 has limits for harmonics depending on the gear.

I've been at a data center where the customer had concerns about harmonics. Power supplies were randomly rebooting and failing before their intended lifespan. The customer accused the data center facility of providing poor power quality. We were able to show that the measured voltage anomalies were due to their switching power supplies and not the UPS or PDUs, but it was still a pain. Even if the IT equipment doesn't care about the harmonics, it is difficult to convince a customer (or potential customer) that this won't be a problem. If equipment is failing at a higher rate, or if we have to rely on warranty to replace something, we may have to provide power quality data.

 

[namesdontmatter]

Otherwise the computers will overheat if you have a power failure. It doesn't do any good to power up the computers if they're going to just shut down due to heat. On the other hand I really think that the HVAC stuff could be put on a generator rather than a UPS but that would seem to be a design decision and up to the engineer who made the decision..

The mechanical selection has a restart time delay that will result in an unacceptable delta T (from which this system may not recover). My vote is to NOT put HVAC on UPSs. Or to have a dedicated mechanical UPS.

 

[petersonra]

Of course they will need cooling. Just not from a UPS. If this is a small installation where maybe there is not a genset usually the UPS just powers the data equipment long enough for a graceful shutdown and the HVAC is not needed to run. On the other hand if this a larger install where the data equipment needs to run longer term there will always be genset and you connect the HVAC to the genset. I have never seen HVAC equipment connected to a UPS as it is an incredible waste of very expensive UPS power.

Pretty much what I said. It seems odd they would use a UPS instead of a generator for the HVAC equipment. But maybe there is a reason that's not obvious. Like there is no natural gas supply available, or there's some local rule against diesel storage. I can't conceive of a situation where this makes a lot of sense but it seems likely that the engineer knows more about it than we do and he is the one that decided to do it this way.

 

[topgone]

Some computer equipment have voltage, frequency and THD requirements. IEEE 519 recommends general systems have no more than 5% THD U. IEC 61000-3-2 has limits for harmonics depending on the gear.

I've been at a data center where the customer had concerns about harmonics. Power supplies were randomly rebooting and failing before their intended lifespan. The customer accused the data center facility of providing poor power quality. We were able to show that the measured voltage anomalies were due to their switching power supplies and not the UPS or PDUs, but it was still a pain. Even if the IT equipment doesn't care about the harmonics, it is difficult to convince a customer (or potential customer) that this won't be a problem. If equipment is failing at a higher rate, or if we have to rely on warranty to replace something, we may have to provide power quality data.

Hmmm. . . The IEEE 519 you have cited recommends a limit of individual harmonics distortion of 5% at utilization voltages below 1000V. The total harmonics distortion limit for any system (THD_v) operating below 1kV is actually 8%. Not picking here, just rectifying misplaced understandings.

 

[namesdontmatter]

Hmmm. . . The IEEE 519 you have cited recommends a limit of individual harmonics distortion of 5% at utilization voltages below 1000V. The total harmonics distortion limit for any system (THD_v) operating below 1kV is actually 8%. Not picking here, just rectifying misplaced understandings.

You're right, thanks, I should have pasted the exact language instead of paraphrasing.

 

[namesdontmatter]

Pretty much what I said. It seems odd they would use a UPS instead of a generator for the HVAC equipment. But maybe there is a reason that's not obvious. Like there is no natural gas supply available, or there's some local rule against diesel storage. I can't conceive of a situation where this makes a lot of sense but it seems likely that the engineer knows more about it than we do and he is the one that decided to do it this way.

There is a generator for the HVAC equipment, but any outage (even ~10s for a generator to start) will result in a 5 minute restart delay for the ACCUs. That delay is unacceptable due to the temperature rise, so that's why there is a push for UPS-backed HVAC.

 

[petersonra]

There is a generator for the HVAC equipment, but any outage (even ~10s for a generator to start) will result in a 5 minute restart delay for the ACCUs. That delay is unacceptable due to the temperature rise, so that's why there is a push for UPS-backed HVAC.

I really misunderstood what you were saying. Probably because I am not a data center designer.

I can understand why they would want to have a UPS that can carry all the loads for a short period of time until the generator gets powered up. One would think that the HVAC loads would be a fraction of the load for the computer equipment.

 

[Jraef]

HVAC can restart after a transition to a generator. I wouldn’t want to risk the entire IT system just to avoid a minute of the HVAC being down. Things don’t overheat that fast. I’ve put an A/C compressor on a UPS for a mobile radar unit, even with a soft starter it occasionally tripped the protection in the UPS and shut the entire radar system down.

 

[winnie]

One would think that the HVAC loads would be a fraction of the load for the computer equipment.

I'd expect the HVAC to be a pretty large fraction of the computer equipment load.

Every kWh of heat produced by the computers needs to be moved by the HVAC system. If the COP of the HVAC system on the 'design day' (worst case conditions designed for) is 2, then it takes 1/2 kWh of electricity to move each kWh of heat.

Jon

 

[petersonra]

I thought these days most air conditioning units were designed to take out about three times as much heat as energy used.

 

[winnie]

You are quite correct about modern air conditioner design. Most air conditioners do better than a COP of 3 _on average_ over the entire cooling season. Something with an SEER of 20 (low for a modern minisplit, dunno what is reasonable for a large data center HVAC system) has an average COP north of 5.

But COP varies with operating conditions, getting lower when the heating load goes up and getting lower when the outside ambient gets hotter.

A COP of 2 would not surprise me at all on the hottest day of the year.

-Jon

 

[paulengr]

I have a data center that wants the UPS to power both IT equipment and HVAC equipment (VFDs/VFDs primarily). I have told them my concerns about having motors and sensitive IT equipment on the same UPS, but they have argued that the VRF systems have soft start, and won't impact the IT equipment. They have done this previously on another build, and I've seen the power quality reports from that build. Highest THD U is ~4%, 5th and 7th harmonics 2-3%. But that building is not fully loaded, so I can't say this won't be a problem in the future.

Does it matter if the HVAC system uses VSD/VFD or not? My intuition tells me any additional non-linear load will create power quality issues.

It has to do with the transformer and the loads. And server power supplies are just as bad. The simple issue is that if all loads were harmonic free with no starting surges transformers are very predictable. A 1,000 kVA transformer can handle 1,000 kVA of load. If we exceed the name plate over the long term depending on outside temperatures and possibly adding fans it just reduces life but overloading by 25% is not uncommon for short periods. As you go above name plate rating though the voltage will begin to drop depending on the %Z. So even though starting a motor can temporarily overload a transformer and slightly dim the lights, it is harmless as long as it isn’t excessive.

Where harmonics cones in is that now we have nonlinear currents and sometimes when the nonlinear effects coincide they can produce a momentary overload of the transformer. This is where current harmonics cause voltage harmonics. Below name plate capacity no matter how ugly and nonlinear currents are frankly we don’t care. It’s when we reach the limits of the transformer and start causing voltage dips that spill over as voltage harmonics that we have a problem.

Voltage harmonics will pass through 6 pulse diode drives putting ripples on the DC bus and can do similar things with PC power supplies. The resulting DC ripples can cause all kinds of effects. Switching power supplies are pretty noisy by themselves. They have significant filtering but it’s not perfect. It can even tolerate some outside noise. IEEE 519 sets an upper limit of 5% on voltage harmonics (THDV).

I’m not sure what THDU is. That’s not a standard term. Might be IEC terminology that uses U instead of V?

So it’s not just harmonics that matters. The transformer (%Z, kVA) also plays a big part in determining what happens. The load and load harmonics matters. ALL 3 phase systems by the way are devoid of harmonics divisible by 3. So you only get 5, 7, 11, 13, 17, 19…single phase systems only produce odd harmonics so with a single phase switching power supply you will see 3, 5, 7, 9, 11…

Cleaning up harmonics from server racks can get ugly because of the 3rd harmonic but often drives can be ordered with either active front ends or DC chokes both of which greatly reduce harmonics. Line reactors work similar and offer protection from surges and short circuits but are more costly. Simple harmonic filters can also do a lot of filtering and they can also add kVARs to improve power factor. Active harmonic filters are sort of a last resort and way overpriced for what they do which is why they are heavily advertised.

 

[GoldDigger]

HVAC can restart after a transition to a generator. I wouldn’t want to risk the entire IT system just to avoid a minute of the HVAC being down. Things don’t overheat that fast. I’ve put an A/C compressor on a UPS for a mobile radar unit, even with a soft starter it occasionally tripped the protection in the UPS and shut the entire radar system down.

When I was at Stanford University in the late 1960s, the campus mainframe computer was an IBM 360/65, in its own building. Intially it had a UPS capable of powering the computer for close to 24 hours. But there was no backup power to the building's A/C. There was a major power failure (car into substation?) that cut off power to the facility.
Despite the UPS, the 360 had to be shut down within an hour to avoid overheating. Unfortunately the mainframe was expected to be powered up continuously, as every cold start (no matter how controlled) burned out several circuit boards, requiring diagnosis and repair.
As a result a steam powered generator was added to the campus heating plant next door to power the computer center A/C only.

On the other side of the balance, the High Energy Physics Lab (HEPL) building had a backed-up circuit to each lab for critical needs (such as power to the mechanical forepump of oil diffusion vacuum pumps.) The only problem was that there was no coordination of its use and eventually the main breaker on the backup power system tripped, causing a guaranteed disaster for all of the labs.