#### Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Options

# Does recycling make sense?

Learning about EROEI from Azimuth has me thinking about whether collecting glass, plastics, aluminium and paper at street collection points makes sense in energy terms? I doubt it, but don't have any convincing reason to say that. I was brought up to believe that aluminium production was energy intensive and had to be sited near hydro-electric power stations. Now you site data centres near cheap electricity!

• Options
1.

You were brought up well! Coal provides around 28 MJ/kg, see table in Wikipedia. Aluminium takes around 54 MJ/kg to produce from ore, see Wikipedia, while recycling takes roughly 20% of the energy.

Comment Source:You were brought up well! Coal provides around 28 MJ/kg, see [table in Wikipedia](http://en.wikipedia.org/wiki/Energy_density). Aluminium takes around 54 MJ/kg to produce from ore, see [Wikipedia](http://en.wikipedia.org/wiki/Aluminium#Production_and_refinement), while recycling takes [roughly 20% of the energy](http://en.wikipedia.org/wiki/Aluminium#Recycling).
• Options
2.
edited October 2010

Carrying on from Graham's point, my vague understanding is that after aluminium and glass most recycling doesn't give any significant energy advantage, if at all, over "creating fresh product". There's other reasons why recycling might be a good thing (reducing pollution from landfills, needing fewer bleaching chemicals for paper and card, etc). It would be good to figure out if it is sensible. Unfortunately, looking around the web all the people criticising recycling seem to be economists criticising it on the basis of a higher monetary cost than landfill rather than on other more physical criteria.

Comment Source:Carrying on from Graham's point, my vague understanding is that after aluminium and glass most recycling doesn't give any significant energy advantage, if at all, over "creating fresh product". There's other reasons why recycling might be a good thing (reducing pollution from landfills, needing fewer bleaching chemicals for paper and card, etc). It would be good to figure out if it is sensible. Unfortunately, looking around the web all the people criticising recycling seem to be economists criticising it on the basis of a higher _monetary cost_ than landfill rather than on other more physical criteria.
• Options
3.

I'll start a stub on Recycling simply to raise some of these questions on the Azimuth Project.

Comment Source:I'll start a stub on [[Recycling]] simply to raise some of these questions on the Azimuth Project.
• Options
4.

Now you site data centres near cheap electricity!

The Register has details of a survey which found almost half of the UK's data centre operators believe the UK's Carbon Reduction Commitment efficiency scheme tax could drive UK data centre investment offshore.

Comment Source:> Now you site data centres near cheap electricity! The Register has details of a survey which found almost half of the UK's data centre operators believe the [UK's Carbon Reduction Commitment efficiency scheme tax could drive UK data centre investment offshore.](http://www.theregister.co.uk/2010/11/03/uk_crc_scheme/)
• Options
5.

On the other hand

Data centre operators have hit back at a British building firm's assertion that high electricity prices in the UK are forcing companies to move their data centres abroad.

computerworlduk

Comment Source:On the other hand > Data centre operators have hit back at a British building firm's assertion that high electricity prices in the UK are forcing companies to move their data centres abroad. [computerworlduk](http://www.computerworlduk.com/news/it-business/16056/no-power-shortage-in-uk-say-datacentre-operators/)
• Options
6.

The other thing, as you're both British, is to remember the "Yes Minister" segment about how to get the answer you want to hear from a survey.

Comment Source:The other thing, as you're both British, is to remember the "Yes Minister" segment about how to get the answer you want to hear from a survey.
• Options
7.

Or we could try some maths. I know little about the economics of data centres, so I shall have to guess some numbers. I guess that if someone is working at a data centre, their share of the data processing is unlikely to use more than 1 kW (while they are working). I can't believe electricity prices vary more than a few pence per unit, so that would only be a few pence per hour, tiny in comparison with wages paid to the worker. Is my estimate of 1 kW way off? If so why?

Comment Source:Or we could try some maths. I know little about the economics of data centres, so I shall have to guess some numbers. I guess that if someone is working at a data centre, their share of the data processing is unlikely to use more than 1 kW (while they are working). I can't believe electricity prices vary more than a few pence per unit, so that would only be a few pence per hour, tiny in comparison with wages paid to the worker. Is my estimate of 1 kW way off? If so why?
• Options
8.
edited November 2010

I still think "could" or "are you concerned" questions are one to which it's very easy to get people to say yes to. If they question had said "will" I'd take it as much. much more definite statement of direction.

Regarding data centres, that value may be reasonable for a scientific data centre where people are chopping and changing batch jobs all the time and there's a lot to be done, but it sounds very low for a future "cloud computing" data centre.

Looking at this info , it's clear that these kind of numbers are a big commercial secret, but here's some kremlinology:

Facebook has over 500 million users and 1 engineer per 1.2 million users, so it has around 417 engineers. It's also stated to have over 60000 servers, so that means 1 engineer is responsible for on average 144 servers. It doesn't say how much energy (for itself+share of network equipment+cooling) a server uses, but I'd expect for anything actually deployed large scale in 2010 a lower bound would be 200W. THis is probably the biggest uncertainty in the estimate: any more supported figures would be welcomed.

So with that lower bound you get 29kW per engineer. Of course. facebook is one of the relatively small number of corporations with that scale of business (to get economies of scale), so you might expect the ratio to drop a bit as the company size decreases. But I see lots of sources saying energy is a more important cost to data-centre's than salaries, and I can believe it. However, latency to customers as well as geographic redundancy (so problems affecting the internet in one area of the world don't render your entire world customers unable to use your product) is also very important is siting. So where a data centre is sited has lots of inputs.

Comment Source:I still think "could" or "are you concerned" questions are one to which it's very easy to get people to say yes to. If they question had said "will" I'd take it as much. much more definite statement of direction. Regarding data centres, that value may be reasonable for a scientific data centre where people are chopping and changing batch jobs all the time and there's a lot to be done, but it sounds very low for a future "cloud computing" data centre. Looking at [this info](http://www.datacenterknowledge.com/the-facebook-data-center-faq/) , it's clear that these kind of numbers are a big commercial secret, but here's some kremlinology: Facebook has over 500 million users and 1 engineer per 1.2 million users, so it has around 417 engineers. It's also stated to have over 60000 servers, so that means 1 engineer is responsible for on average 144 servers. It doesn't say how much energy (for itself+share of network equipment+cooling) a server uses, but I'd expect for anything actually deployed large scale in 2010 a lower bound would be 200W. **THis is probably the biggest uncertainty in the estimate: any more supported figures would be welcomed.** So with that lower bound you get 29kW per engineer. Of course. facebook is one of the relatively small number of corporations with that scale of business (to get economies of scale), so you might expect the ratio to drop a bit as the company size decreases. But I see lots of sources saying energy is a more important cost to data-centre's than salaries, and I can believe it. However, latency to customers as well as geographic redundancy (so problems affecting the internet in one area of the world don't render your entire world customers unable to use your product) is also very important is siting. So where a data centre is sited has lots of inputs.