Thermodynamic calculations for a server room

[letsbangout]

Hi all. Thanks in advance for looking; this forum is a great resource! I operate several small server rooms and I'm looking for guidance as I try to perform a calculation that I haven't seen referenced anywhere (yet).

Right now I have three water-cooled heat pumps in this particular room which are 2.5 tons each. They're placed at the end of my server room, which is rectangular and approximately 25' long and 16' wide. The ceiling height is kind of hard to estimate. At its highest point it's about 11', the bottom of the steel beams are 10', and there are all kinds of impediments to airflow (racks, servers, poles for the ladder rack, the ladder racks themselves, fiber trays, steel electrical and fiber conduits, water pipes, etc). The 2.5 ton units are upflow and the nameplate says that the blower motors are 0.25HP (is that right??). The heat pumps are capable of getting down to 69 degrees right now, using a cooling cycle of approximately 5 minutes in duration. The thermostats are set to kick off for a mandatory 5 minutes before re-engaging. So when they cycle on again after the 5 minute timer, they're usually at 75 degrees, sometimes 76. Other vital information about this room - it has approximately 135 amps of AC power in use by the equipment, along with another 8 overhead fluorescent lights (they're the kind with two bulbs each, and look to be 4ft. long each). Under typical conditions, only one person is present in the room at any given time. Also maybe of potential interest is that the ambient temperature outside of the server room is typically 76 degrees.

What I'm trying to determine is how many more watts of heat I can add in the immediate proximity of the ACs (i..e within 4 or 5 feet) before they can no longer cool within reasonable parameters. When I say reasonable, I mean a reasonable duty cycle on the compressors, whatever that may be. Perhaps 20 minutes on, 5 minutes off? How much heat can I add and still reach perhaps 73 degrees at the thermostat after a 20 minute cooling cycle? What about 74 degrees? Or 75? What are these calculations called?

Let me preface this also by stating that I already have a handle on the basic watts to BTU/hr conversion, etc. As such, my calculations tell me that these ACs can remove 26.390KW worth of heat (90000 BTU/hr * 0.293 watts per BTU/hr) - I don't know how practical such a calculation is, or at what point i'll run out of cooling short of the 26.390KW figure. When it comes to the air-flow part, I know less than nothing, so any help is appreciated. Thanks again!

[flange]

You really need to do a complete heat gain for the room, as you have missed a few things. First, you are considering the total btu capability of three 2.5 ton units to be 90000 btuh. Which theoretically is correct, but you must remember that most server rooms are considered high sensible loads, and very low latent loads. In other words, you have a need for mostly temperature change not dehumidification. A standard heatpump wil have a total rating close to what you surmise, but it needs to be broken out as sensible and latent, so you may not really have as much capacity as you think. For example, one heatpump that i looked at with a 2.5 ton rating had a total cooling capability of 29200 btuh, of which it could do 21300 btuh of sensible cooling. There are other factors to be considered as well such as ambient conditions during building unoccupied periods. There would be a greater load when the surrounding areas are not cooled as they are during occupied periods.

[temco]

Ditto on What Flange said on the heat load. Sounds like a mechanical eng. ,local contractor with in house eng. and experience with server rooms would be a good call.
Also a server room and its loads are a different animal than an occupancy load especially with new servers with concentrated heat loads. Using the wall thermostat satisfying or not wont confirm that you equipment is being cooled properly. Hard for me to say what you have going without a lot more info. But when you flood a server room with cooling like you would your typical office this doesnt necessarily mean your server is receiving the cooling it needs. You can provide 20 tons of cooling to a server room that only needs 12 tons and still have your equip. starving for cooling because most of your cooling is by passing your load.

[chillrdude]

The answers to all your questions are here:
http://www.hvac-talk.com/vbb/showthr...ghlight=server

[letsbangout]

You really need to do a complete heat gain for the room, as you have missed a few things.

Like what? I've named everything that produces heat and indicated specific quantities of each - overhead lighting, server equipment, and humans. Also forgive my ignorance as I don't know what "do a complete heat gain" means.

First, you are considering the total btu capability of three 2.5 ton units to be 90000 btuh. Which theoretically is correct, but you must remember that most server rooms are considered high sensible loads, and very low latent loads. In other words, you have a need for mostly temperature change not dehumidification. A standard heatpump wil have a total rating close to what you surmise, but it needs to be broken out as sensible and latent, so you may not really have as much capacity as you think. For example, one heatpump that i looked at with a 2.5 ton rating had a total cooling capability of 29200 btuh, of which it could do 21300 btuh of sensible cooling.

That's the kind of thing I was looking for. The little bit of googling I've done on the definitions of sensible and latent cooling would support your premise. You're exactly right that I need to effect temperature change, not dehumidification. In fact, contrary to popular belief, humidity is completely irrelevant to this equipment and its effect in terms of failures is not significant enough to be quantified or acknowledged. That said, how can I determine the real-world sensible cooling capability of these heat pumps?

There are other factors to be considered as well such as ambient conditions during building unoccupied periods. There would be a greater load when the surrounding areas are not cooled as they are during occupied periods.

76 degrees is accurate on a 24/7 basis.

[letsbangout]

Also a server room and its loads are a different animal than an occupancy load especially with new servers with concentrated heat loads. Using the wall thermostat satisfying or not wont confirm that you equipment is being cooled properly.

I'll address that independently; the focus of my question is simply on removing heat.

[letsbangout]

The answers to all your questions are here:
http://www.hvac-talk.com/vbb/showthr...ghlight=server

Thanks. I've read over the first couple of pages; I'm not sure I have the patience to read all 51. I should be ok if I can figure the sensible cooling capacity of these heat pumps.

[temco]

When it comes to the air-flow part, I know less than nothing, so any help is appreciated. Thanks again!

I guess I was responding to this, and that you mentioned that there was all kinds of impedements to airflow. Your air flow patterns and ability to transfer conditioned air directly to your servers before it has mixed with recirculating hot air from the servers is something (among others) that might be considered in addition to the heat load calc.
If your server requires 200 cfm of 57deg air and you are providing it with 200 cfm of 65 deg air ??

[letsbangout]

I guess I was responding to this,

I was referring to the way that air-flow relates to my calculation in the original post. But you're making some very good points.

and that you mentioned that there was all kinds of impedements to airflow. Your air flow patterns and ability to transfer conditioned air directly to your servers before it has mixed with recirculating hot air from the servers is something (among others) that might be considered in addition to the heat load calc.
If your server requires 200 cfm of 57deg air and you are providing it with 200 cfm of 65 deg air ??

Being the manufacturer of the machines in question, I'm pretty familiar with what they need in terms of temperature and CFM. The fact is that by design, I assume that internally they'll build up to 115 degrees and receive 5 CFM across each machine, being cooled mostly by the flow of air across the front and back of their chassis (and real-world measurement confirms that). As previously mentioned, I'm concerned only with my ability to remove heat from the room. Hot-spots can be solved by installing an exhaust duct that couples fully or partially to the intake of the heat pumps - unless I'm missing something. But the exhausting/ducting/etc. seems easy compared to the science behind cooling itself.

[flange]

The equipment manufacturers will have a section in their equipment guide that lists the total, sensible and latent heat values for a given system. you need to find the contractor or distributor of the equipment to get that. If you post it here in a different thread, someone might have it handy. As far as total heat gain calcs, does your rom have any expsures to areas that are of a different temp? That could affect the rom heat gain.