Hacker News

Embodied energy of digital technology(lowtechmagazine.com)

102 pointsfemto posted 5 months ago50 Comments
l0b0 said 5 months ago:

The "source" is just a list of almost 200 publications, with no indication of which numbers come from where. So it's effectively unsourced. That means it's either bullshit, terrible reporting, or both, and there's no reason to believe this over i-have-an-agenda.org.

I've got karma to burn, so bring it. But it really is time that reporters get their bloody act together. Without the research this is as worthless as an opinion piece, and it's not going to convince anybody who needs convincing.

amirhirsch said 5 months ago:

The fun part is seeing those ranges for silicon after having spent the last several weeks researching ways to decrease the energy costs of producing “solar grade” (one nine less than monocrystalline wafer) silicon from silica. There’s a paper showing 1.6kg of 6-nines silicon from rice hull ash in 6hours in a 50kw arc reactor (300kwh) (https://pubs.rsc.org/en/content/articlelanding/2015/gc/c5gc0...)

rosser said 5 months ago:

> Furthermore, burning rice hulls to produce electricity and RHA, generates more energy than required for the overall process.

I don't have access to the paper, but assuming I'm reading that correctly, the notion of burning rice hulls to power the process and thereby source their ash is a fantastic efficiency.

londons_explore said 5 months ago:

That lies between "silicon" and "electronic grade silicon" in the posted table, as you'd expect.

Rice hulls cost something to produce too, and I bet you need a lot more than a kilo of them to make a kilo of silicon...

dredmorbius said 5 months ago:

The sourcing could definitely be better. The values smell about right to me (I've been exploring this space for a few years), and spot checks suggest they're reasonable.

Alternative sourcing: https://www.sciencedirect.com/topics/engineering/embodied-en...

(Hits: timber, recycled and stainless steel, aluminium, glass, plastics.)

Enough so that I'd be inclined to believe the values. Low Tech Magazine is generally reliable in my experience.

l0b0 said 5 months ago:

You do realize that's several logical fallacies in one?

"smell about right"

"spot checks suggest they're reasonable"

"generally reliable in my experience"

dredmorbius said 5 months ago:

You may have failed to note the alternative sourcing link provided in my comment above. Or my top-level comment linking ICE, in this thread.

stewbrew said 5 months ago:

The list is somewhat biased:

    Iron (from iron ore)
    Steel (from iron)
    Paper (from standing timber)
Is the energy for digging up the ore or felling that tree + transport + machinery for production included? Steel should probably also include the numbers for iron.

The list would benefit from a common baseline. Without that these are just random numbers.

londons_explore said 5 months ago:

I'd prefer it said:

Steel (from the things Steel is typically made from in 2019 - 80% iron ore, 50% process X, 20% recycled)

That gets tricky, because many industrial processes use large amounts of energy, but only because that energy is effectively 'free' as waste heat from other processes.

dredmorbius said 5 months ago:

The list is specifying clearly what the derived energy values are for. Not that they're necessarily typical (though I suspect these are).

There's a tremendous variability to industrial processes and material utilisation, including virgin vs. recycled materials. Listing from among options may help clarify engineering / manufacturing / consumption values and choices.

kensai said 5 months ago:

The original article is also interesting: https://www.lowtechmagazine.com/2009/06/embodied-energy-of-d...

"The 180 watt laptop

While these reports are in themselves reason for concern, they hugely underestimate the energy use of electronic equipment. To start with, electricity consumption does not equal energy consumption. In the US, utility stations have an average efficiency of about 35 percent. If a laptop is said to consume 60 watt-hours of electricity, it consumes almost three times as much energy (around 180 watt-hour, or 648 kilojoules).

So, let's start by multiplying all figures by 3 and we get a more realistic image of the energy consumption of our electronic equipment. Another thing that is too easily forgotten, is the energy use of the infrastructure that supports many technologies; most notably the mobile phone network and the internet (which consists of server farms, routers, switches, optical equipment and the like)."

londons_explore said 5 months ago:

Gas turbine powerplants are easily in the 60% + efficiency band... Coal at 30% is dying out in most of the western world.

Granted every fossil fueled device ever has quoted efficiency with the Lower Heating Value, whereas the HHV is the fairest measure. The difference consists of if you should extract energy from the humidity in the exhaust, and the clear answer is yes.

dredmorbius said 5 months ago:

AFAIR "easily" is a stretch -- there are highly efficient combined-cycle gas-turbine plants, and I think some of those actually get above 50% efficiency.

For most thermal generation, however, it's Carnot-cycle limits which get you down, and that's governed by hot vs. cold side efficiency.

Coal actually has some of the higher efficiencies, in some cases, exceeding 45%, though whether that's through high temps or combined cycle I'm not sure. (I'm not defending coal, just noting efficiency numbers I've encountered.)

Update: Wikipedia's language is "possibly 62%" efficiency, FWIW: https://en.wikipedia.org/wiki/Combined_cycle_power_plant

londons_explore said 5 months ago:

You're right - only the best ones achieve 60+% efficiency.

For example, [1] achieved 63.08% efficiency at the whole-site level (ie. including inefficiencies in startup, testing, etc. over a year)

[1]: http://interfaxenergy.com/article/30230/chubu-electric-plant...

dredmorbius said 5 months ago:


dasd99 said 5 months ago:

> Another thing that is too easily forgotten, is the energy use of the infrastructure that supports many technologies; most notably the mobile phone network and the internet (which consists of server farms, routers, switches, optical equipment and the like)

While I agree with you, it's worth noting that the server farms of our tech giants are already powered by renewable energy (Google 100%, AWS 50%, Microsoft 50%, ...).

mikestew said 5 months ago:

I'm going to have to cut down on my titanium addiction. OTOH, my nephew's grandchildren will be fighting over that double-walled coffee mug long after I die.

jsilence said 5 months ago:

Gather the good stuff while energy is still cheap!

Waterluvian said 5 months ago:

Recycle your aluminum, people!

femto said 5 months ago:

This paper also adds information on absolute energy usage and per kilogram produced.


Gibbon1 said 5 months ago:

I've found this paper tantalizing.


""High purity iron was produced, with a current yield of 85% and a power consumption of 4.25 kWh/kg iron.""

If you run numbers you get energy costs of about $400-$500/ton.

edejong said 5 months ago:

Would be nice to contrast this with the energy necessary to produce 1 kilogram of various foodstuffs.

dvh said 5 months ago:

Is electronic grade silicon created by chemical vapor deposition? I thought czochralski process.

BostonEnginerd said 5 months ago:

It's both, I believe. Highly purified polysilicon is created using the Siemens CVD process. This is then broken up and used as feedstock for a CZ crystal puller.

namibj said 5 months ago:

Neither. It's a zone meelt technique that uses the fact that the solubility of impurities is higher in a more liquid state of silicon.

dredmorbius said 5 months ago:

For those seeking sources for embodied energy, ICE seem to be among the more rigorous and accepted sources (I've encountered them before).

Wikiedia cites similar values referencing ICE.



quixoticelixer- said 5 months ago:

Does the value from wood come from processing it or does it include the energy the tree used to make it?

rebuilder said 5 months ago:

It says "From standing timber", so I'd say that's the energy used in processing, whatever that is defined as.

JKCalhoun said 5 months ago:

I understand we put aluminum recycling plants near rivers (like the Columbia) where energy costs is cheap.

deepsun said 5 months ago:

Or the other way around. In USSR, at least, few power plants were built closer to aluminum ore deposits.

Gibbon1 said 5 months ago:

Yeah three really strong covalent bonds takes a lot of energy to break. And strong enough you can't use carbon alone to reduce it. Basically it's bottled electricity.

hollerith said 5 months ago:

I heard the same about the plants that refine the aluminum from ore. Also that those plants being there was a major reason a large airplane manufacturer (Boeing) is nearby.

ahigherugliness said 5 months ago:

Is measuring in kilograms a good way to think about this? How many kilograms of chips does my phone have?

dredmorbius said 5 months ago:

It normalises the metric across all levels of production. Scale according to use.

A 5" iPhone weighs 138g. Much of that is probably battery, display, and case, so chips are on the order of 10% or less of the total. I'd estimate at <10g.

The core processor is the A10 Fusion SoC, 125mm^2, probably a few mm thick. Silicon has a density of 2.3290 g/cm^3. That adds up to about 20-30g, depending on thickness, though I'm not sure of the density or thickness of a silicon wafer. I suspect actual mass is less.

philipkglass said 5 months ago:

Numbers are not sourced and, in at least some cases, suspiciously high. Virgin aluminum is more like 14 kWh/kg, not 63.


8bitsrule said 5 months ago:

The OP-linked page points to a source URL at the bottom (An MIT 'Environmentally Benign Manufacturing' bibliography).


After that URL it points back to a Low-Tech page called 'The monster footprint of digital technology'


For a summary graphic, see figure 2 in this cited PDF (Gutowski et.al)


philipkglass said 5 months ago:

I did refer to figure 4 in that PDF. It shows aluminum around 16 kWh/kg for smelting. According to the accompanying pie chart, electricity used in smelting accounts for 75% of the energy used in aluminum production while bauxite mining, alumina refining, primary casting, and anode production account for another 25%. Starting from 16 kWh for smelting that would put the full energy intensity at 21 kWh per kg of aluminum.

8bitsrule said 5 months ago:

Yeah, I misread the Figure numbering in that paper, it should have been '2' (which has a big black '4' next to it).

On the whole, and prima facie, that graphic delivers much the same message, and rings true. Can't expect the L-T author to do -all- the work.

hollerith said 5 months ago:

Maybe 14 kWh/kg is just the electricity used in the refining process.

Maybe to get the 63 kWh/kg figure, you add to the 14 kWh/kg figure the energy needed to collect the cans (if recycled) and to mine the aluminum (if not) and the energy needed to build the refining or recyling plant and the energy used by the workers who mine, collect and operate the refining and recycling plants, including the energy used to feed, clothe, educate, entertain, transport, etc, the workers.

In other words, maybe one figure attempts to estimate the second- and third-order energy use, i.e., the "total carbon footprint", and the other does not.

benj111 said 5 months ago:

"including the energy used to feed, clothe, educate, entertain, transport"

As Sagan might not have Said:

If you want to make a kilo of virgin aluminium, you must first invent the global economy.

ryan_j_naughton said 5 months ago:

This made my night and brought back some very dear memories. Thank you.

Paying it forward, here is the original clip so others can get a hit of the sweet sweet nostalgia...


dredmorbius said 5 months ago:
philipkglass said 5 months ago:

That page does not include any of the numbers in the Low Tech Magazine table. It lists 165 academic and 14 popular publications.

The other paper posted by femto shows aluminum at ~16 kWh/kg. The slightly higher value is due to an earlier date cutoff around 2009. See figure 4(b):


dredmorbius said 5 months ago:

Fair point, though it's also not untirely unsourced. I'd agree to "poorly".

mrpurple_ said 5 months ago:

What about fuels like petrol? :)

roberto said 5 months ago:

A plot here would go a long way.

chronolitus said 5 months ago:

Here is a quickly made plot, both linear and log-yaxis. https://i.imgur.com/BIRAZzJ.png

dredmorbius said 5 months ago:

And my versions of the plot, log and linear axis: