If nothing is wasted in a chicken processing plant, why not data centers?
Modern poultry processing maximizes the yield of human food from each animal. Chicken livers, hearts and giblets are sold with whole birds or separately. Chicken feet are common in Asian supermarkets. Scraps are sent for processing into animal feed. Manure is often used as fertilizer either in raw or composted form. Feathers may not be economical to process locally and are often disposed of by incineration although I believe I’ve found one use in the pokey feather pillows my wife bought at the big brand low-priced decorating store. Despite this, many articles point to research for chicken feather uses including raw materials for plastics and as an additive to concrete to improve strength.
If this is true for chickens, why not look for uses of waste from data centers, waste heat in particular? After all, we pay good money to heat our homes and water, so why not use a waste source for the same purposes? In an earlier piece I looked at Cogeneration or CHP, wherein a gas turbine is used to both produce electricity by powering a generator and the heat is used either for heating systems, hot water heating or as a heat source for a steam powered generator. While CHP can be very cost-effective for data center power, this piece looks at using the low grade heat from a data center’s HVAC system in heating buildings.
The idea is not new, but is not necessarily widely used. A 2011 research paper published jointly by Microsoft and the University of Virginia (Link to Source) discussed data centers as home heating systems, which they coined as Data Furnaces. The idea was picked up by The New York Times (Link to Source) in which the author discussed putting servers in homes for a massively distributed cloud data center. Location in the home is needed to reduce heat loss during transmission. While this may not be practical (imagine data center technicians knocking on your door at midnight when a server rack goes down) the paper notes the idea is economically sound resulting in an average savings of approximately $300 per server per year across all heating zones.
Data Furnace table shows average $300 per server per year savings vs. conventional DC.
The paper notes the primary challenges to the idea of home based server racks, primarily isolation making monitoring challenging, security, and what is termed zero-touch management, meaning the servers can continue to operate at some level until service can be scheduled. These are not small concerns and would be difficult to manage in rural Montana, but that may not be true in high-density urban areas.
Data Center Knowledge posted a 2011 piece titled Energy Efficiency Guide: Heat Recycling (Link to Source), that looks at several existing and planned projects that can be thought of as Data Furnaces. Among those discussed:
Finland's Uspenski Cathedral (left) is above heats hot water which will then be piped to nearby homes for heating. Treehugger.com’s data center (Link to Source) Waste heat from servers at the Telecity Paris data center heats an on-site arboretum (right)
In Uitikon, Switzerland outside of Zurich the waste heat from a data center built by IBM for GIB Services AG will heat a nearby swimming pool (below). (Link to Source)
As in the case of CHP systems, concentration of the waste data center heat to make it more suitable for heating air or water may be needed. This is generally done by employing a heat exchanger, a heat pump in many cases. Heat equivalent of the electrical energy used is multiplied up to seven fold in modern systems making them very cost effective. Alternate approaches are being explored such as work by Quantacool where microchannel heat exchangers are distributed throughout a server rack to extract heat from the hottest parts of the server can be captured more efficiently making sure servers run as cool as possible.
Work by Professor Alfonso Ortega at The University of Villanova and others is being developed into commercial systems at Quantacool Corp. (Link to Source)
We’ve all been told if something looks too good to be true it probably is. The same is true of using waste data center heat for home and commercial space and water heating applications. Like PV solar, methane fuel cells, wind, battery storage and other green or renewable technologies there is a cost both in equipment (HVAC ducting, air-to-air and air-to-water heat exchangers, control systems, etc.) and the real estate to install such devices. Not all data centers are candidates, but those that are may be well served to explore the RoI and potential green publicity which may be more valuable in the long run. Of course, this requires a long run view, so in the end that may be the biggest obstacle of all.