X
Business

Seattle looks to teach the new datacenters an old trick

Reusing waste heat in a way that benefits the community
Written by David Chernicoff, Contributor
Teaching an old datacenter new tricks

According to an article in last week’s Economist, the city of Seattle is looking to get a major benefit from something that the current crop of the city's new datacenters is throwing away — waste heat. By making use of an old distributed heating technique, Seattle's Office of Sustainability and the Environment thinks that what is now a problem that datacenters need to deal with can be turned into a positive that benefits the datacenter operator and local community alike.

Dealing with the heated water, which is the result of cooling the datacenter IT workload, can be a major pain for a datacenter operator. Recirculating the water through a series of devices to first heat it up, then cool it down can be a complex, and expensive process.  Few datacenters have the luxury of an unlimited supply of chilled water and a way to easily dispose of it, so most are faced with the costs associated with getting the water, then dealing with it once it is heated.

Seattle is suggesting that datacenters become part of what is called district heating, where water is heated at a central location, then pumped to a network of buildings that use the water as a primary heating source. This isn’t a new idea and is commonly used in many cities in Europe. I covered how it was being used by a datacenter in Helsinki, Finland about three years ago.  It used to be a fairly common practice in the US, but fell out of favor as different forms of heating became commonly available.  It is still used in some locations, though, and in Seattle a company called Seattle Steam still heats some 200 buildings using this technique.

The trick, of course, would be building the infrastructure to distribute the heated water. On its own, the water isn’t quite hot enough to be used as a direct heat source, so a company like Seattle Steam would need to be an intermediary, taking the pre-heated water and raising its temperature further. This would, of course, reduce their cost of generating the water at the proper temperature, but it means that the infrastructure needs to be in place — not just to heat facilities from the water, but to deliver the water from the datacenter to a central location for additional heating.

Upfront costs would look to be a significant issue; conversion of heating from electric to water radiators in existing buildings might be problematical, though new construction could be built with this heating scheme in mind. And the datacenter operators would need to figure out how to route their cooling and what if anything, they would charge to provide this warm water to the steam utility.

This plan has worked in a few existing locations in Europe primarily because the heating infrastructure was already in place and the technique is a common part of the public consciousness. Here in the US though, the startup costs, compared to the return on the investment back to the public, are likely to be insurmountable in any major urban location.

Editorial standards