Data Center

Examining the Amazon Plan for Re-using Data Center Waste Water

I have always believed that innovation is a team sport. So I read with interest the recent blog post about how Amazon is planning to use waste heat from a nearby data center to heat a new office complex it’s building in Seattle. The post starts with a legitimate point:

“The idea makes so much sense that it prompts one to wonder why no one thought of this before.”

But the issue isn’t really that nobody has thought of this idea before. As the post makes clear, the concept amounts to a hydronic heating system, where warm water is used to supply heat; hardly a novel idea. But taking that warm water – technically, heat media or thermally conductive liquid – from a data center, that’s the new and tricky part.

Simply put, we haven’t seen many examples of this (if any) because it’s hard to do. It takes a lot of planning and engineering to pull it off, mainly because the data center waste heat media contains low-level heat from the IT equipment in the data center. The heat that comes out of a server is typically 270F (150C) higher than inlet temperature, in the vicinity of 900 – 1000 F. In some blade server installations the temperature rise may be as much as 420F, with an output temperature as high as 1100F degrees.

Using that low level heat to warm up a building requires it be moved out of the data center and transported to wherever it’s going to be used, all without radiating away too much of the absorbed heat in the process. That means you need insulated piping to ferry the thermally conductive liquid and some sort of large, insulated cistern to hold it when it gets to its destination – which better not be too far away or you will surely radiate and lose most of the heat. (First Law of Thermodynamics, gotta love it.)

You also need a favorable climate. During winter, the idea makes a lot of sense in most climates, and perhaps spring and fall. But clearly northern climates – like Seattle – are more apt to use the approach.

There’s also a pretty intricate balancing act that has to happen in order to produce a temperature that’s comfortable for humans. If you simply moved 100-degree heat into a building, that would clearly be too warm for most of us. So you need to blend it with some outside air in order to achieve the comfort sweet spot of 70 degrees or so. To do that you need building controllers on each side, one taking heat out of the data center and the other putting heat in to achieve a temperature that’s acceptable for humans. It’s a balancing act that requires fairly deft compute work on both sides.

If you can get it right, though, it amounts to a big win. The way we cool data centers is to take all the low-level heat from inside the data center, put it into thermally conductive liquid, transport it outside and evaporate the heat into the atmosphere. In the process, we lose two things: the electricity it takes to move the heat from point A to B, and the water that winds up getting evaporated into the atmosphere through chillers.

Amazon and its data center partner, by figuring out how to make use of that heated data center waste water, in effect get a two-fer. Amazon gets reduced electrical or gas costs for heating its building while the data center conserves water.

I expect we’re reaching the tipping point where we’re going to be seeing more of this kind of approach for making good use of waste heat from data centers. It’s already a constant dialogue I hear when traveling to Europe, which in general has a cooler climate than here in the U.S. (and especially where I live, in Texas). Data center managers there are always wondering what they can do with the heat their facilities produce.

I can picture some kind of centralized waste heat storage system that data centers feed into and various buildings pull heat from, with a metering system to figure out how much everyone pays or is credited. We are now at the point where the technology is there to make such a system feasible and cost-effective. And if cities and/or power companies provide financing or other incentives for companies to build such systems, as I suspect they will, all the better. Kudos to Amazon and Seattle for their innovations and effort, and especially for demonstrating it can be done.


No Responses

Leave a Reply

  • (will not be published)