[Beowulf] Electricity cost: a critical survival issue of our ICT infrastructures.
Lux, Jim (337C)
james.p.lux at jpl.nasa.gov
Thu Apr 11 12:14:41 PDT 2013
I like the idea of solar panels to power your cluster.. no energy storage required, just run when the sun shines. Yes, there is a substantial ecological/waste problem with solar panel manufacture.. so maybe we need to scale up and run solar thermal for our cluster. You can build a nice 5-10 MW solar thermal plant, and that should run a few thousand nodes quite nicely. Electricity costs are estimated at around $50-60/MWhr from parabolic trough or power tower designs.
But to address Vincent's concerns about data sources and such:
The data comes from the US Government Energy Information Administration which provides independent statistics and analysis. It's pretty complete, detailed, and cross referenced with original sources. The EIA doesn't really have much of an axe to grind: they're more about collecting data, unlike, say, DoE (whose origins as the Atomic Energy Commission always make it slightly suspect for me)
Nuclear power is decreasing in the US because all the nuclear plants are getting old and being decommissioned and there's essentially no new plants coming on line. No new plants started since 1974 and Three Mile Island in 1979 basically put paid to any new starts and led to lots of cancellations. The other problem was the huge financial overruns on plant construction (WPPS). Unless they're heavily subsidized, it's hard to make money with a nuclear plant, and that doesn't even begin to get into the costs of waste disposal and storage, for which there is no long term solution in the US. Today, wastes tend to be stored at the point of generation.
We will just wait for fusion power, too cheap to meter, from seawater which is (always) 20 years in the future (and has been since 1950). (and we'll worry about the radioactively hot waste stream then.. it's not fission products that are the problem.. neutron activation is a big issue)
The big safety issue with nuclear power is the long lasting effects of a unlikely but severe disaster. Unlike a hydro dam break (which has comparable probability and immediate impact), a serious nuclear accident has effects that literally last millennia. Chernobyl, Fukushima, TMI, etc. all have BIG downstream effects which will last substantially beyond the prompt effects.
Coal usage is not surprising. There's an enormous coal mining, distribution, and consumption infrastructure (e.g. the big coal fired plant in the southwest which basically on top of a coal mine). Coal is cheap, particularly for surface mining. The cost is basically digging it out of the ground and dumping it into rail cars for transportation. And it's found in a lot of places.
However, the remediation costs for emissions and such is becoming prohibitively expensive. Inevitably, there's "other stuff" along with the carbon in coal, and burning gigatons of it is a good way to extract those substances (Sulfur, Mercury, etc.). It would be exceedingly expensive to build a brand new coal fired plant and meet emissions rules. To date, what they do is a succession of "minor modifications" (which can be surprisingly large in scope) to an existing older plant that doesn't have to meet the present rules.
From: Vincent Diepeveen [mailto:diep at xs4all.nl]
Sent: Thursday, April 11, 2013 11:09 AM
To: Eugen Leitl
Cc: Lux, Jim (337C); Beowulf at beowulf.org
Subject: Re: [Beowulf] Electricity cost: a critical survival issue of our ICT infrastructures.
At european energy market one has to PAY if you deliver windmill or solar panel energy at hours the market doesn't need the energy.
So a huge energy producer exploiting such windmill parks effectively loses more money than he makes. Same with solar plants - yet there is very few of those and the ones there are very tiny.
This is a realistic worldview on solar and windmill energy of course and should be also translated into not subsidizing it except for research.
Right now it's moving money directly from EU to China which produces windmills and solar panels, of course at a price no one here can compete with.
Gas always has been huge in Netherlands as energy source (we also export gas) yet oil hardly gets used to produce electricity.
World wide usage of coals only is increasing, especially in 3d world nations. The nations using lots of coals only use more and more of the stuff.
I'm a bit amazed though USA is using that much coals and gas as when compared to the nuclear production of it.
Of course assuming the table is correct that Jim quoted, as there is no market where that much desinformation gets spreaded as in the energy world, as there is always billions at stake and politicians simply don't want to bring the bad news to their voters that there is as of yet no good way to produce electricity without some sort of major disadvantage, be it CO2 output or killing all fish and eco life of a river with a hydro plant or the risk of a melt down of a
nuclear central. They just are not prepared to tell the truth and governments then mess up statistics not seldom.
On Apr 11, 2013, at 12:59 PM, Eugen Leitl wrote:
> On Wed, Apr 10, 2013 at 09:40:38PM +0000, Lux, Jim (337C) wrote:
>> In the US, electricity comes mostly from coal and natural gas, with
>> the latter rapidly replacing the former. France is somewhat unusual
>> in having significant nuclear generation, but in the US, nuclear has
>> been roughly constant at about 20%.
> It doesn't have to be nuclear, coal or gas
>> In PetaWattHr
>> Coal 1.517
>> Gas 1.231 (natural)
>> Nuc 0.769
>> Hydro 0.277
>> Renew 0.219 (wind, tidal, solar)
>> Oil 0.013
>> Other 0.012 (no idea what this is, biomass?)
>> Gas 0.011 (other, blast furnace gas, e.g.)
>> Coke 0.010 (from oil)
>> Over the last few years, Coal is decreasing by about 200 TeraWh/ yr,
>> Nat gas increasing by about the same. Oil is decreasing by about 3-4
>> TWh/yr, renewable is increasing about 20-25 TWh/yr.
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin
> Computing To change your subscription (digest mode or unsubscribe)
> visit http://www.beowulf.org/mailman/listinfo/beowulf
More information about the Beowulf