[geeks] Ideal Humidity range for server rooms?

jodys at helluin.org jodys at helluin.org
Wed May 6 09:49:12 CDT 2009


On Wed, May 06, 2009 at 08:43:52AM -0400, Bill Blum wrote:
> We've had some.... 'issues' with the A/C unit in our server room
> recently.    Four times in past few months we've been alerted to temps in
> excess of 90F/32C, and the humidity has gone as high as 70%.   ( Building
> HVAC and plumbing FAIL. )
> 
> 
> It's a small room ( approx 400 sq ft), so our single AC unit should in
> theory fit the bill.     When it's *working*, it'll easily keep the ambient
> air at 75F, and the area behind the server racks at about 78F.  Problem is,
> the humidifier/dehumidifier functionality is somewhat problematic--- and for
> bonus points, my boss is convinced anything about 35% RH is pure, total
> evil.
> 
> Are there any whitepapers out there that have 'ideal' ranges for data
> center/server room relative humidity???   I need something I can point to
> and say that it's not so bad SO TURN OFF THE EMAIL ALERTS ABOUT 36% HUMIDITY
> ALREADY.

ASHRAE recently modified there reccomendations for datacenters. They now
use dew point, in addition to relative humidity. Here is a short report
on the new parameters.

http://shorterlink.org/5183

I think the relevant (to high humidity) quote is

"Based on extensive reliability testing of Printed Circuit Board (PCB)
laminate materials, has been shown that conductive anodic filament
(CAF) growth is strongly related to relative humidity [3]. As humidity
increases, time to failure rapidly decreases. Extended periods of relative
humidity exceeding 60% can result in failures, especially given the
reduced conductor to conductor spacings common in many designs today."

The situation with low humidity is slightly less clear, but related to
ESD. It appears that low humidity is correlated with increased
discharge, but "discussions with the IT equipment manufacturers
indicated that there have been no known reported ESD issues within 
the current recommended environmental limits." So maybe the parameters
are overly cautious.

My understanding is that the tight humidity ranges are a holdover from having
big line/chain/other-exotic-types-I'm-not-familiar-with printers in the
data center. But don't hold me to that.

As a datum, our facility has a setpoint of 40% RH, though most of the
units seem to hover around 35%-36% RH at their sensor (wasting power...)
and we haven't had excess failures.

Jody



More information about the geeks mailing list