Recycling Excess Heat From the Data Center

itwbennett writes "A new data center being built in Helsinki, scheduled to go live at the end of January, will generate energy and deliver hot water for the city. The data center is located in an old bomb shelter and is connected to the Helsinki public energy company's district heating system, which works by pumping boiling water through a system of pipes to households in Helsinki. The recycled heat from the data center could add about 1 percent to the total energy generated by the energy company's system in the summer." The article doesn't say what the overall efficiency of the heat recovery is. Researchers at MIT are working on a new energy-conversion technology based on quantum dots that they say has already demonstrated 40% of the Carnot efficiency limit — 4 times what is achieved by current commercial thermoelectric devices. The researchers believe they can reach 90% of the Carnot limit.

Re:I think it's great, but...

[onionman]

Is pumping boiling water through pipes the most efficient way to heat houses? Isn't there a pretty massive heat loss in the pipes?

Having said that, if they are already using this system for heat, the introduction of waste heat from a datacenter seems to make a lot of sense. Acts as a heat sink for the data center, reduces the amount of energy needed to heat the water.

Actually, it's a reasonable system for heating large building complexes where a central facility can heat the water. Many Universities and large corporate complexes already use similar methods for heating their campuses. Insulation on the pipes keeps the water warm in transit. These complexes can also cool their buildings in the summer by pumping chilled water through the system.

Re:Data center cooling in the winter

[afidel]

That's correct, in many climates you would spend more energy trying to maintain humidity then you would save by not running the AC system. For larger systems many have a "freecool" option which runs the waste stream through the heat exchanger without running the compressors thus saving ~80% of the energy.

Re:Valium and Xanax for Engineers and Physicists

[JessGras]

Heat pumps can be this efficient when you consider the claim is they can move five times as much (heat) energy as they consume (in electicity or other organized source). They are not claiming to generate more than they consume: only pump more than they consume - though vague and sensationalist phrasing in the journalism makes this unclear in the OP.

It IS great

[dgr73]

Not to mention that when you have hot water coming out of the pipes, you don't need a water boiler, which is something all houses without kaukolämpö (remote heat) need. All in all, the infrastructure is in place in many places in Finland, with insulated pipes dug deep enough into the ground to keep the heat, so why not take advantage of it.

Re:WTF, why is a Carnot reference here?

[vlm]

How is the Carnot cycle apply here? This is direct heat conversion, and the efficiency should be near 100%, you would have line losses.

Reading thru a filter of journalist ignorance, I think the journalist is trying to say they're using a heat pump to reject the heat from the cooling system into the heating system's boiling water. There is a Carnot cycle limit to heat pumps.

Re:Data center cooling in the winter

[djrok212]

This already exists, and is in use in many data centers, they are called Air Side Economizers, http://www.42u.com/cooling/air-side-economizers.htm [42u.com]

Forced air is too dry

[mi]

Well, they have to be, because you're basically threading a steam bomb throughout your entire house. I'll stick with forced air thank you.

Forced air will dry you into a raisin. It is December — do you notice, how dry your lips are in the morning?

You need humidifiers to fight that effect... No, hot water — pumped through fixtures made of cast iron, or something, that's even slower to warm up and cool down — is the best heating solution... It could be expensive, but it is the most comfortable of what's commonly available today.

The oft-used copper and/or aluminum fixtures are bad, because the temperature will be fluctuating widely between the times, the heat is turned on by your thermostat and the times, when it is off. Our bodies are more sensitive to changes in temperatures, than to the temperatures themselves. Also, a quickly-heating material ends up losing heat mostly through convection (heating up air, that rises to the ceiling), than through radiation, which warms you directly (via infra-red).

Stainless steel is better in that regard than copper/aluminum, but not as good as cast iron, heavy and "unattractive" as those things might be...

Re:WTF, why is a Carnot reference here?

[jank1887]

because the summary is talking about two entirely separate topics. (1) Helsinki data center will deliver waste heat in useful form to the city. (2)Researchers at MIT are working on a solid state heat-to-electric conversion element called a thermoelectric device. current devices are at best ~10% of Carnot (practical devices approach 10% total efficiency on a good day), and they say they'll hit 40-90% of carnot with their new quantum dot TEG's. We'll see.

You are correct, though. Carnot efficiency discussion only really applies to (2) not (1). Mister itwbennett just decided to lump these two things into one submission.

Re:WTF, why is a Carnot reference here?

[Snufu]

The summary writer may also have wanted to suggest that thermoelectric devices may be used to convert the waste heat at the data center into useful energy. However, thermoelectric devices are not really feasible for this application. To achieve the efficiencies cited for thermoelectric technologies, you need temperature gradients in the 100's degrees C, as in, for example, a car engine. Typical temperature gradients at data centers are on the order of tens of degrees C. This is considered "low-grade" heat in the renewable energy vernacular.