PDA

View Full Version : Computer Rooms



vspiteri
20-01-2002, 08:48 PM
What are the parameters for a very large industrial computer room in eliminating area hot spots

Brian_UK
20-01-2002, 10:39 PM
It depends on how the cold air is to be distributed. Is it from ceiling mounted diffusers (I hope not) or floor mounted grilles (my favourite) ?

If you are using separate aircon units (wall or ceiling) then I think that you should investigate the expected airflow patterns from them and position them accordingly. There may be a requirement for additional air circulation fans but of course this adds to the heat load.

Additional information on the room type, equipment selection would help.

Gibson
21-01-2002, 05:09 AM
What Brian said! More info is required to answer your question vspiteri.

Most computer rooms here use downflow units with floor distribution tiles that are easily moved to accomodate the loads around the room.

vspiteri
21-01-2002, 09:06 PM
Brian & Gibson

Yes, the air conditioned air is down flow type through several AC units.

The floor tiles near the equipment are perforated , the temp and humidity are maintained 21 to 55% respectively within the room ,but the outlet of the AC units varies from 08 to 12C.

Gibson
22-01-2002, 05:50 AM
vispiteri,
It would be helpful to know the A/C unit manufacturer.

55% RH sounds high for a computer room set point. I'd check with the computer room manager to see if I could drop the set point to 45% RH.

I'd ask for a plan of the computer room that includes the installed loads.

Next I'd sketch the configuration of the perforated floor tiles on the plan, obtain the tile manufacturers airflow ratings of the perforated tiles.

Most computer room a/c units I work on are designed for 600 cfm/ton. See if there are enough tiles to match the requirements of the a/c manufacturer.

In some cases such as you describe, if there are hot spots, there are cold spots. Sometimes a redistribution of the tiles is all it takes. Matching the airflow through the floor tiles to the heat loads.

In some situations, more tiles are required. Some applications call for more cooling.

vspiteri
25-01-2002, 09:22 PM
Thank you for the feedback.

The parameters that I need are:

Sensible heat ratio -

Load density (m2/ton) or (ft2/ton)

Thanks

Gibson
26-01-2002, 06:30 AM
vispiteri...
I think I gave you more credit than you deserve. Were you diagnosing a problem in an existing computer room or designing a new one?

Load density? Sensible heat ratio? Another test! *LOL*

Seems to me that you must be studying out of a book with a test at the end of each chapter? You forget to include the other pertinent information from the chapter and expect us to hit on the correct answers?

Could you use "load density" and "sensible heat ratio", in a few sentences to show us how they might solve your problem?

Might the terms have anything to do with airflow and distribution?

How does airflow through a computer room A/C affect the sensible heat ratio?

What happens to the sensible heat ratio if there aren't enough perforated tiles to match the airflow design of downflow computer room A/C units?

What happens to the sensible heat ratio if the perforated tiles are placed too close to the unit? If the humidity set point is too high?

What do you call the other kind of heat?

Yes, this is a test! *S*

vspiteri
26-01-2002, 09:54 PM
Gibson

I explain myself on this computer room,since I didn't want to make long stories and it is going to be one.

The computer room,this is an industrial room and I know that running around with tiles which perfectly done and maintanied one can solve the the normal problems ,but here the problem is much more problamatic.

These large computers are testing units that test computer chips,which at high RH and Temperature will stop testing the devices or failing them ( worst ) due to a parameter increase while after few seconds this computer will fail and it will take a lot of time to regain their performance - it's a matter of process than.

The technical problem is that under the plenium the RH will be 55% - 60% and the temperature 10C (outlet of the evaporator fan) . The airflow outlet from the floor tiles going directly to these test machines (computers) which have the electronic cards at low level, very close to the floor plenium .

Along the process at 1 meter height in the computer itself, the RH will decrease from 55% to 38% while the dry bulb temperature will increase to 20C ,all this due to the load,at the test machine outlet(high level) the RH will be below 20%(creates defects) and the dry bulb up to 30C or more.

It is here that the whole mixture of these different loads from test machines of any type (high and low RH's) that will began effecting the system while the area is governed approx by 320 tons of refrigeration (details will be given if interested although it is very complex).

At this stage , while inspecting and doing the root cause, I always notice plenty of hot and cold spots at the floor area and at 1 meter above floor level while 0.5 meter below false ceiling is an inferno,which is the return of the split unit system which has to maintain the recirculating air at the requested parameters ,

It is here that am concentrating on all imaginary results and parameters that are not normal parameters and that they effect the system,maybe a matter of density that these large machines have within their occupied space (hardware) and the electronic density(power boards ) within the machine itself due to RH separation within it's height from bottom level to the top level.

Please do note that as Plant I cannot increase the RH at low from the plenium because it will block the machine due to high RH (tollerance +/- 5%) or lower the dry bulb temperature to 8C due to operations health requirements - operators which at 1 meter above floor plenium must be 19 to 23C.

This is a mixture of industrial load and human comfort.

That is the reason why I need all these technical parameters so I can have enough proof to consolidate new ideas to the suppliers and top managment which as for sure you know and is aware that they need scientific results (numbers and facts).

My problem is that I have to answer to the top management by numbers , facts and no assumptions since the investments are very high while payback periods must be very short ,not more than 2 years - corporate targets.

Gibson,I am not testing but I need facts and something to began working on,if I will have these parameters I will ask for proof and from where I can obtain an original document so that the ball will began rolling -

I never though that you were going to to use the same attitude as some of the forum techs.

I didn't want to insult nobody and am very sorry if I did,but we are all humans and retaliation is one of our defects,it is a matter of how one will retaliate and what massage was delivered.

Noticing that you are criticized by tech who have only seen 1/2" pistons will drive you mad - as most probably my answer to rotation has done the same effect to you - we are all humans- the problem is that the forum is global and not individual.

I do hope that this has cleared your things.

Thank you

Victor

Gary
26-01-2002, 10:59 PM
Given these symptoms, I would suspect insufficient airflow through the computers, and/or the A/C units.

What are the return air and supply air temperatures for each A/C unit?

Has the system worked properly in the past? If so, what has changed?

Square foot (or square meter) per ton is not an adequate way to determine BTU requirements as the majority of the load presumably comes from the computers. Such an application needs a full and accurate heat load calculation.

vspiteri
26-01-2002, 11:17 PM
Gary

Thanks-

Today I cannot give flow details, but the computers are all fixed with HEPA filters which do have high resistance to the airflow and to overcome flow problems they have individual ( 2 in number) large high flow extractor fans on their higher level facing the false ceiling - outlet air flow.

This gives them forced air across every computer or test heads.Sorry that I didn't mentioned them before but things will began on detail with detailed questions are asked.

Victor

Gary
26-01-2002, 11:27 PM
You might try removing the HEPA filters temporarily to see if performance improves.

Also make sure the extractor fans are clean and running in the right direction.

Are these squirrel cage type fans or propellor type fans?

Jasper
26-01-2002, 11:28 PM
Vspeteri

It seems that one of the problems that you have with this project is stratification of air in the higher levels of the room.
This could be overcome by installing slow moving fans at high level:rolleyes: :p

vspiteri
26-01-2002, 11:54 PM
Gary

Yes I agree with you that the main problem is load but as you are aware the electronics industry shifts equipment on daily basis and this will not permit a detailed calculation.

These days the market is very low and changes which costs are high and no return is seen are cancelled- best operation

I was thinking of installing a 120000 m3/H AHU at the facility perimiter and installing flexibile ducting from the main branch which will have the ability to be moved and connected to the floor plenium as supply ducts and above the false ceiling for return air.This will reduce the direct load from the recirculated air which is the direct mean to the test heads.

The matter is that I need energy conservation principles (free cooling) AHU at a point that the recirculated air which is costly treated by the split units would not be wasted and while outside air will be used on frequent basis.

This is investment and hard planning and feedback is needed,that is why am trying to solve the problem isolating on a small scale by single units to have the proper feed back from the system before large investments will be proposed or though of or maybe having an idea of other large facilities.

Thank you

Gary
27-01-2002, 12:01 AM
The extractor fans would be more effective if they discharged directly into the return. Next best would be to have return grilles directly above the extractors. The discharge from these may be bouncing off the ceiling, causing hot spots.

Mick, I suspect that such fans may be counter productive in a floor supply/ceiling return application. Good thought, though.

Gary
27-01-2002, 12:27 AM
When considering air flow problems, keep in mind that return grille size and placement are every bit as important as supply outlet size and placement. It is all part of the air flow strategy. The best place for return grilles is closest to the heat sources.

Outdoor air is generally considered very dirty for a computer room. In many areas it may be against local code.

Also, if you are going to introduce outdoor air you must exhaust air to the outdoors or the room will be excessively pressurized.

Both exhaust air and makeup (outdoor) air should be installed in the return flow, with air exhausted upstream from the makeup air. If you are going to do this, it may make sense to use an air to air heat exchanger.

Either way, the effectiveness of the existing system should be maximized.

vspiteri
27-01-2002, 12:27 AM
Gary /Mick

Thanks

Hepa filters - Yes I changed these filters with new ones because am not allowed to removed them and work without them since they make part of the process machinery - an other department.

To this reason I suggested that a dp meter will be installed to all HEPA while the pre-filters will be cleaned on daily basis

matter on large scale persisted

The extractor fans are squirrel gage and propeller type and their rotation is OK.

Stratification may be the problem but what type of slow moving fans we may suggest when the total height of the complex is 3 meters - ceiling fans must be installed more that 7 ft above floor.

I would like to add too that most of these test heads have a refrigeration chiller to cool their head - this chiller is air cooled so the condenser is pumping hot treated air in the air conditioned space which the cost is high to throw away since the exhausted air must be replenished with outside air - suggestion was done to change these air cooled to water cooled chiller as a main unit with many cooling circuits to increase redunduncy on break down while it will be installed outside of the facility. -

Thank you

Victor

vspiteri
27-01-2002, 12:41 AM
Gary

Thanks - For developing

What would be your proposal and how can I use the hot return air from the individual computers if I run it (hot air flow) through a recovery wheel and transport it to the inlet of the AC split unit.

I need to recover every single cent befor I propose projects.

victor

vspiteri
27-01-2002, 12:55 AM
Gary/Mick

For today I have to stop because down here it is 01.00 hrs in the morning / night.

Victor

Gary
27-01-2002, 01:07 AM
Computer rooms are not my specialty, although I have worked on many over the years. Perhaps sufficient information has been drawn out for others to jump in and offer their expert advice now. I feel we've made some progress. And tomorrow is another day. :)

vspiteri
27-01-2002, 09:17 AM
Gary & Mick

Thanks

victor

dannyd
26-02-2002, 11:51 PM
I assume that the existing a/c units are all located along the perimeter of the room?, if so can you advise if all the cooling stages are on, or are some cooling circuits not energised?. This will give some indication of the overall spare capacity - if any exists. Advise on the method of air return to the a/c units, does air return freely to the top of the a/c units, and if so does the temperature vary greatly from unit to unit?.
It may help to consider if a ceiling void exists,then can you fit return air plenums to the top of the a/c units and (if ceiling construction permits) fit return air grilles in the ceiling to produce a 'laminar flow' effect, to better mix the high temperature air from the load with floor supply air. This may assist in eliminating the high discharge air temperatures from your load which seems to hang below the ceiling level.
I presume that the slighly high r/h levels are a requirement of the computer manufacturer, bear in mind that the higher r/h levels requested can add to the room load as the humidifiers steam merrily away - check how much of this is being stripped back out on the evaporator coils.
Hopefully, not too many questions have been asked at one time, but I would be interested in your reply