The myth of the 15-ton compressor

For those of you that don’t know I started in the industry as a facilities technician, then within three months became a site lead mechanical engineer. One thing that always perplexed me was the persistence of the 15-ton (or any nominal sized refrigeration component) compressor. Even today, working as a regional engineer over 3000 miles of data center footprint, I still deal with this from people who should know better.

So what is he griping about now? Well the rating on equipment is a nominal rating. It is the equipments performance AT a specified set of conditions. Calling a compressor by it’s nominal rating and acting like that is the exact value of cooling it can always provide is a dangerous path. But again and again, from HVAC technicians, to sales engineers, to data center design engineers, I have heard things like that:

“No, those are only 15 ton compressors”
“No, that’s only a 350 ton chiller”

First, let me approach the possible scary side for most people operating data centers: in the real world, at your real world conditions, the performance could actually be much lower than the nominal value. Applying a 15 ton compressor in an rtu is great, but it’s capacity may be impacted by the evaporator coil design, the condenser coil design and ambient considerations, the space settings you are trying to maintain and the return air conditions.

Let’s talk about some anecdotes to drive that point home.

I have seen a center where they started using another manufacturer for their scroll compressors because they were cheaper. While innocent enough, they were impacting the very design of the entire unit and the 3 other compressors which run in the same circuit. This cheaper compressor was also lower in efficiency and capacity, and while it was still rated 15-ton nominal, the impacts of the change are far greater than a true “like for like” replacement.

In another center, I have seen air cooled chillers placed close to each other (albeit within the manufacturers minimum clearance requirements), surrounded by a 20 ft and a 40 ft wall on 3 sides, where the nominal performance was far from realized. On a 95 degree day, 4 350 ton chillers were all running at 100+ % of their rated load amps and cooling an approximate 900 ton load. How in the world, if those are truly 350 ton chillers, would that math work?

It doesn’t, and that’s because this all circles around the operating conditions. Those chillers were making 42-40 degree water, slightly below the design condition, but air re circulation meant that they were essentially operating as if we were in Dubai with 135 degree dry bulb temperatures, as that was the temperature of the bulk of the air entering the chiller.

So that was the “People are dying!” headline, but there’s a less fanatical converse argument. The 15 ton compressor in that rtu, well it’s not just 15 tons. As data centers have evolved and ASHRAE recommendations for computer room equipment are accepted, the typical inlet has come up to around 75* or more, and the implementation of good containment practices have elevated return temperatures to 90* or more.

Side note, get friendly with your vendors when you push r-22 or r-407a past a 90* return as they may have not tested that equipment for those levels and the suction gas temps may be inadequate to cool the compressor. Find out what the real limit is, but be pushy about it as the vendor will try to give you an overly conservative answer.

Back on track, changing an rtu from a design condition delivering 52* air and handling a 75* return temp to a 75* supply will have huge impacts on the capacity and effectiveness of its cooling. To help with some general thumb rules ( which you can see proven out on Manufacturer’s equipment where they list multiple capacities) please remember these two:

An increase in return and supply temps of 1* f will increase efficiency and capacity by 2-3 percent.

A decrease in ambient temperature of 1* f will increase efficiency and capacity by 2-3 percent.

These rules start to fail at extremes but hold mostly true for the ranges of cooling used in data centers.

So in the above, let’s say our rtu was designed for a 95* day and we want to change that design to 105 to comply with Tier III requirements, but we also have revised our sa temp from 52* to 72*. The net of this change is about a 20-30 percent increase in efficiency and capacity of the compressor. Sound like a lot? Then check out some package equipment specifications and see how much a ten degree change actually does. Liebert posts good data, and for their new DSE line they have been very forthcoming with publishing data for higher return temperatures. Off the top of my head, a DSE 150 with 75 degree return air is like 135 kW of cooling, at 85* return is 150 kW, and at 95* is 173 kW. Remember, the DSE is a highly optimized piece of equipment, and it’s engineers have already squeezed out a lot of the extra capacity. (Bravo Zulu to Emerson on those bad boys).

As far as ambient conditions go, Liebert assumes that you size your heat rejection for the design conditions, so that means they don’t have to discuss it when talking about the package unit.

Point being, you don’t buy a car that gets 25 MPG and assume that is its fuel efficiency under any conditions. If you drive to work on windy hills or in the city, you can expect different efficiencies than if you have a straight road with no traffic lights. Some conditions may be better and some may be worse. Keep the same mindset when considering your cooling equipment. It performs at a certain level under certain conditions. It may be better or worse based on the real world conditions. That same 350 ton chiller might be making 280 tons, or 410 tons, depending on your water temps and ambient temps. As with many things, the devil is in the details.


One thought on “The myth of the 15-ton compressor”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s