Hard Drive Failures

[eastcreek]

List,

The electrical installers at this site put all 42 current carrying conductors into one under slab pipe to supply power to the computers in our call center.
I realize harmonics can be a problem when non-linear loads are powered. My main question is, could our too frequent hard drive failures be occurring because extra harmonics are being added due to the installation? I can go to any outlet and measure 4-10 volts between the ground and neutral.
I also realize a catastrophic failure could occur here if there is a melt down of the wiring under the slab or if the 60% loaded dry type transformer fails.
Any help is appreciated.
eastcreek

 

[dereckbc]

Re: Hard Drive Failures

Eastcreek, first is the system a 3-phase 208/120? I assume it is. 4 volts from neutral-to-ground is a result of the load current on the neutral, and not the result of harmonics.

If you have 4 volts dropped on your neutral, you can probably assume you have 4 to 8 volts being dropped on phase conductors, which means you are more than likely about 108 to 112 at the outlets. A little low IMO. You would have to measure to be sure.

Harmonics are greatly over exagerated. The biggest danger from harmonics is overheating the transformer, and neutral conductors which is easily solved by uasing "K" rated transformers and oversized neutral conductors.

Really need more info to answer, but I doubt harmonics are causing your hard-drives to crash. Sounds to me like your branch conductors or feeders are either overloaded, too long, or both.

Voltage drop design for these type of installations should be kept to 3.3% or less. 2 volts or less between N-G, or 4 volts or less P-N.

[ January 21, 2004, 05:32 PM: Message edited by: dereckbc ]

 

[brian john]

Re: Hard Drive Failures

Depending on the load on these 42 current carrying conductors....and the size of the conductors (did the contractor derate the conductors?) I would be concerned about the failure of the insulation and the resulting shut down of your call center. The harmonics can add to the heatinmg in the conduit as noted.

 

[tonyi]

Re: Hard Drive Failures

Electrical possibilities aside, is there any pattern to the failures?

IBM had a batch of 60G IDE drives a while ago that came from a plant in Hungary that are known to have attrocious failure rates. The same model from other plants were OK. (you'll find these bad drives in lots of OEM gear, not just IBM machines)

Look at where the machines that are failing were physically placed. Machines on the floor under desks tend to get kicked a lot.

What operating systems are being run? There are vulnerabilities in the bootstrap loaders for Win95/98/ME and DOS, that will allow someone to sabotage a machine in such a way that it won't boot even from a virgin boot floppy. The symptom when this is done mimics a hardware failure. A quick check on this possibility is to attempt to boot a copy of IBM's PC DOS 7 on the machine. IBM fixed this vulnerability in that DOS release (Microsoft never fixed it in their versions)

 

[eastcreek]

Re: Hard Drive Failures

List,

The contractor came out of the panel with #10 wire and ended with #12 wire at the receptacles.
The panel is a 120/208v. In my mind the contractor did NOT derate...if he had, I think he would have put 10 amp breakers on these circuits instead of the 20 amp square D bolt-in's he used.
310-15 mandates a 35% value derate for 41 or more conductors. 240.4 limits the maximum current for a #10 wire used for computer circuits at 30 amps. 35% of 30 is 10.5.

On the circuits I see the 4-8 volts neutral to ground (this value fluctuates)I see the 120 fluctuate also. Sometimes dropping to 115 volts.

Per an article in the February 2000 EC&M Power Quality Advisor written by a "professional engineer", "When servicing a high concentration of nonlinear loads, power distribution systems can experience a wide variety of problems, such as: Misoperation or component failure in PLC's, computers, or other sensitive loads."

The plant uses all Dell CPU's. Other sites in the area are using the Dell's also without the frequent failures we see here.

All that said, is it a proven fact that wire to wire induced emf does occur? If yes, in my mind...this could add to voltage/current distortion in a circuit. Can anyone substantiate this?
eastcreek

 

[dereckbc]

Re: Hard Drive Failures

Originally posted by eastcreek:
List,

The contractor came out of the panel with #10 wire and ended with #12 wire at the receptacles.
The panel is a 120/208v. In my mind the contractor did NOT derate...if he had, I think he would have put 10 amp breakers on these circuits instead of the 20 amp square D bolt-in's he used.
310-15 mandates a 35% value de rate for 41 or more conductors. 240.4 limits the maximum current for a #10 wire used for computer circuits at 30 amps. 35% of 30 is 10.5.

You are quoting the correct sections of the code, but your interpretation is incorrect. You de-rate the conductors, not the breaker. 20-amp breakers are the correct size for the branch circuits. It is the conductor that sounds undersized. IF there were only 3-current carrying conductors, #12 is all that is required by code. Since you say you have 41 conductors you have to use 310.16 to find the conductor. Do not have my code book handy right now, but I guess a # 6 AWG would be required for the application.

None the less, it sounds like your problem is a more of a design issue, along with a code problem (conductors not de-rated). How long are the branch circuits from the distribution panel? Is there a PDU and UPS used? If all you are using is a dry type transformer that is not "K" rated , multi-wire neutral circuits, long branch circuit distribution, and no line conditioner, then you have a design issue.

It is true that if you have more than 2-volts from N-G you can run into problems either being harmonic or too much voltage drop. The solution is the same, oversize the distribution conductors to overcome voltage drop, use separate neutrals for each branch circuit, use "K" rated transformers to handle the extra heat from harmonics if a UPS is not used, or use UPS and PDU's to clean up and provide constant clean power. Finnaly balance the load to minimize neutral current and harmonics.

[ January 22, 2004, 10:46 AM: Message edited by: dereckbc ]

 

[eastcreek]

Re: Hard Drive Failures

dereckbc,

Thanks for the reply.

I do realize that derating is normally used to determine wire size. I mentioned I felt the contractor did not derate due to his 1st mistake which was to small an underground pipe for 42 conductors and 14 neutrals. (ieeeee...what a mess!) So....all he was left with would be to downsize, or derate, his breakers (which was not done).
Branch circuits from this 3 phase panel are approximately 250'. There are no PDU's or UPS's or even a K-Rated transformer. (I think our saving grace may be that it appears to be only 1/2 loaded.) Reinstallation is not an option at this point.
My concern is getting my report into the powers that be here. I've told them that we definitely have some electrical problems and that I felt a good part of them are being caused by the installation.
I intend to check for load balance. I'll take a neutral current rating before I begin. I wasn't aware load imbalance can cause problems also.
Would still like to know if anyone can verify or negate my theory of cross emf occurring in the underground pipe which I think could add to the harmonic problems that exist here.
eastcreek

 

[tonyi]

Re: Hard Drive Failures

With 250' runs, voltage drop will be an issue too.

 

[iwire]

Re: Hard Drive Failures

No one has asked what the loads on these circuits are.

Many times each circuit in a occupancy like this run individual pieces of equipment, it would be likely that each of these circuits has less than 5 amps of load.

At this level of loading voltage drop even at 250' would be less than 5% and we would be under the derated capacity of the conductors.

To meet the NEC the use of 10 amp breakers or 15 amp breakers if the branch circuits feed single receptacles (240.4(B) would be required.

I am not saying this is a great situation or that it presently meets the NEC, I just like to look at things in a more positive perspective.

As a side note here in MA those 42 current carrying conductors would derate to 18 amps instead of 10.5.

 

[luke warmwater]

Re: Hard Drive Failures

Are you looking at any other possibility of these failures than a bad electrical design or electrical problem?

Are you in a high lightning area?

Who has replaced the hard drives and what were their findings?

What is the running temperature of these computers? Or are they servers?

Did the main system get infected by a virus?

There are many things that could be causing these hard drives to crash. Not just electrical problems.

 

[brian john]

Re: Hard Drive Failures

To add to LW questions are there any power supply failures in these machines? Or just hard dirve losses?

 

[pwhite]

Re: Hard Drive Failures

eastcreek,
you may want to consider a constant voltage transformer. they protect against voltage spikes and brownouts. sola is the brand we use, i'm sure there are others. they are expensive though as compared to an electrically shielded transformer. we also use battery backups.
pwhite

 

[charlie tuna]

Re: Hard Drive Failures

this installation has too many problems to be able to pin it down to one! first and formost when supplying computer equipment "dedicated neutral conductors should be used" - read the instructions on the equipment's box when you open it - "this equipment is designed to be connected to a dedicated circuit"! common neutrals used on computerized equipment cause hormonics within that single neutral conductor. this is multiplied by the problems of "voltage drop"(250 foot branch circuit) and conductor deration (42 current carrying conductors and 14 neutrals in one raceway). harmonics cause "heat" problems in source equipment like transformers, but they also cause problems in computers because they don't allow currents to mix on a common neutral.

years ago i bid a sizable job for the tenant "goldman sacks" - a large stock exchange firm. the entire tenant space floor was raised. the plans called for 1 1/4" subfeed raceways run under the floor and then distributed to different desks and workstations. the number of circuits shown on the plan, along with the neutrals, indicated to me that they were "common neutrals". example: 12 circuits(hots) and 4 neutrals and a ground in each 1 1/4" raceway. the engineering firm "was" one of the biggest in the area. at the first jobsite meeting i approched the engineer and asked him if this was "in error" - between the two of us not in front of the other attendees. he said "it's my job to put it on paper and it's your job to install it"!!! wow! we installed it as shown. about two years later i was called to attend a meeting due to all the major problems they were having-equipment failures-lost data - damaged lazer printers-etc.. and who was running the meeting-yep- the same engineer! i had my original "as built" prints. he brought another set - a new set - his set had "dedicated circuits" with two inch raceways. his had over sized branch circuit conductors to compensate for the voltage drop. we started into the meeting and he rolled his set out and started telling everyone that the "installation was wrong" and that it would require major repairs and that these repairs would require overtime to prevent interfearing with normal office operations. would you say this guy had a set of big ones!!! goldman sachs asked me what i was going to do??? i told them i was probibly going to make everything right---AT THE ENGINEERS EXPENSE! i rolled out my "as built" drawing and compared dates - the dates were the same as his drawings! you could planly see these were "as builts" and then i explained my conversation with this same engineer at the pre jobsite meeting. he wanted to take my drawings-"there must be some mistake" he said! i told him "if goldman sachs asked i would provide them a copy of my drawings but for financial purposes these drawings would remain in my hands". goldman sachs had us correct the job - and we did the corrections the way we wanted to without engineered drawings and during normal working hours and were pair by goldman sachs - i don't know if they were re-imburst by the engineering firm?

these were the same problems:
1) no dedicated circuits
2) no consideration of derating conductors
3) no consideration for voltage drop

on computerized equipment!!!!!

 

[dan55]

Re: Hard Drive Failures

You must have about a 2in.feeder pipe going to your call center, would it be possible to put a sub-panel in the call center then you could re-feed with 3/0.

 

[eastcreek]

Re: Hard Drive Failures

List,

I've just completed some checks on this panel over here.
36 amps neutral (high?)
27-39 phase a
19-39 phase b
17-29 phase c
No circuit is pulling more than 6.3 amps

I put a cheap frequency meter on the conductors and found that 10 of the circuits don't constantly read 60 hz. Some were reading as high as 180 hz. This somewhat answers my original question about induced emf effecting other bundled conductors. I suppose since not every conductor exibited odd frequencies this may not be occurring in a pronounce way (osicliscope might prove otherwise)like I thought it could.
I do realize that keeping 3 phase circuit condutors together is the common practice...still not positive how 42 kept together behave.

To answer some of the posted questions:

1)Some lightning in the area, however what's more frequent is temporary 2-10 second brownouts which seem to occur once every 3-6 months.
Note:The computer guys usually loose one or 2 when this happens. (They have just informed me that it is the power supply that fails 50% of the time after the brown outs.)(I'm not saying hard drives and or power supplies aren't failing during utility provided power....which looks clean.)

2)The ambient temperature around the computers are 73 degrees. (Any cooler in the room and I'm the first to hear about it!)

3)No viruses present according to the network guys.

I like the idea of using the 2" pipe to feed a subpanel closer to the loads. I think this would be cheaper than running 6 or 8 conduits up 20' and out across a busy production floor to the call center 200' away.

Bottom line is, as Charlie has eluded too, when computerized equipment is being installed, keep the installation to code and run a neutral for every phase conductor. (Then keep your fingers crossed?)

eastcreek

 

[dereckbc]

Re: Hard Drive Failures

Originally posted by eastcreek:
I like the idea of using the 2" pipe to feed a subpanel closer to the loads. I think this would be cheaper than running 6 or 8 conduits up 20' and out across a busy production floor to the call center 200' away.

Your best solution is to move the transformer where the power is utilized. If that cannot be done then you could add a sub-panel, but you would need to base the feeder size to voltage drop (1-2%) rather than OCPD.

 

[charlie tuna]

Re: Hard Drive Failures

easty,
the code does not require separate neutral conductors on three phase circuits when suppying computer equipment. common sense does! i am not suprised at the readings you recorded concerning full load currents. this indicates harmonics are present and to a heavy extreme! remember the computer is looking for a source voltage of 120 volts. when the voltage instantaniously creeps up on the neutral conductor due to the circuit's inability to mix and cancell each other out--- the voltage to "that" particular computer my be 85 volts - at that point operating currents throughout the computer's printed circuit boards is near twice normal - it's almost like hittting the computer with a hammer! how many "hammer hits" will your hard drive take? the first indication usually is that some data is lost? you will find the neutral current reading down around three or four amps when dedicated circuits with proper sized conductors are used. the harmonics will then be cancelled out on the neutral block in the distribution panel... i'm sure due to induction - some real funny stuff is happening within that 2 inch pipe full of branch circuits! and be aware - the computers probibly are damaged from being exposed to this situation, so don't expect lost hard drives to stop!!!

 

[dereckbc]

Re: Hard Drive Failures

Eastcreek, this may sound sarcastic, I do not mean it that way, but it may help sell your management.

I do a lot of data center design/build projects, and here are some questions you might ask.

How important is the data/IT/phone systems to the operations of daily business?

How much down-time is acceptable due to PQ problems.

How much has outages and down times cost in the past three years?

From what you describe the data/IT/phone systems are not important, any amount of downtime is acceptable, and outages have not cost them anything. Right?

You have got yourself some serious design issues, and a band aid is not going to solve the problem.

See if one of the accountants can estimate the losses due to lost data, damaged equipment, labor, etc over the last three year period.

At a minimum you should have a special isolation transformer called a PDU installed near where the power is going to be used. And if brown-outs and power outages cannot be tolerated, then a dual conversion UPS should be considered to feed the PDU.

Good Luck