RE: Has anyone looked into using the power model presented in this paper to for estimating co2 emissions for data transfer?

Hi Chris,

We faced similar limitations when building our impact model, and of course the paper you mentioned felt very relevant to us. All sources regarding network impacts are expressed in energy per data transmitted, or in impact per data transmitted when the manufacturing impact is accounted for. We tried to dig deeper by accounting for the different impacts of the different network types (copper, fiber, 3/4/5G).

In particular, some countries provided the network mix between copper and fiber and between generations of mobile network, allowing us to account for it through the location (country) of users. Some of our costumers are also able to report the share of user relying on home network or on mobile networks. Otherwise, we have to provide results for usages scenarios (Wifi vs. mobile network).

Regarding the usage impact of network (ie, converting energy to environmental impact), we also account for the location of the network : if the users and datacenters are in different country, an average electricity mix for this country can be used to convert the energy to environmental impacts. Finally, we account for the hardware involved in the network  (eg, the end-user's router or additional network equipement), through their power usage and a depreciation of their manufacturing impact over time. However, this is essentially accounting, and the estimated impacts will depend on the accounting rules that are used.

We wrote a blog article last year providing a more comprehensive explanation [1]. We also recently opened our impact model in a pair-reviewed conference [2]. This paper introduces all our hypothesis, formulas, and uncertainty management for impacts on end-user's devices, network  (section IV.B), and backends.

We would be glad to hear about how others do this kind of calculations !

[1] https://greenspector.com/en/what-impact-does-the-network-have-on-digital-services/
[2] https://hal.science/hal-04532041/document

Thierry LEBOUCQ
Président GREENSPECTOR
+ 33 6 88 71 50 50
GREENSPECTOR – 6 rue Rose Dieng-Kuntz 44300 NANTES
www.greenspector.com<http://greenspector.com/> | @Green_spector<https://twitter.com/Green_spector>
________________________________
De : Chris Adams <chris@thegreenwebfoundation.org>
Envoyé : lundi 8 juillet 2024 17:19
À : public-sustyweb@w3.org <public-sustyweb@w3.org>
Objet : Has anyone looked into using the power model presented in this paper to for estimating co2 emissions for data transfer?

Hi folks,

An interesting paper was published late last month that proposed a new way to talk about the environmental impact of data transfer.

It’s called Network energy use not directly proportional to data volume: The power model approach for more reliable network energy consumption calculations, and is published as an open access paper at the link below

https://onlinelibrary.wiley.com/doi/10.1111/jiec.13512

Here’s the abstract:

> It is commonly assumed that data volume and network energy consumption are directly proportional, a notion perpetuated by numerous studies and media coverage. This paper challenges this assumption, offering a comprehensive examination of network operations to explain why the relationship between energy consumption and data volume is nonlinear. The power model approach is explored as an alternative methodology for calculating network energy consumption providing a more reliable representation of network energy use. The power model demonstrates that simple energy intensity calculations, expressed as kilowatt hours per gigabyte of data, are insufficient for accurately estimating real-world network energy consumption.

The lead author David Mytton, also writes a more accessible summary on his Substack, and he’s been working on it since 2022.
https://www.devsustainability.com/p/network-energy-use-not-directly-proportional

I've created an issue for anyone who fancies thinking through how it might be implemented below:
https://github.com/thegreenwebfoundation/co2.js/issues/218

If you have looked into it, and how you might turn it into a model you would use in a some software, I’d be interested in hearing about it, as I think this is novel, interesting, and welcome work.

At the same time, it’s not as familiar and easy to use as the SWD model we have as a default inside CO2.js, the library we maintain at the Green Web Foundation, so I think the discussion would be welcome - especially if it helps us develop better mental models for thinking about digital sustainability.

Thanks!


Chris Adams

Executive Director

w: thegreenwebfoundation.org
e: chris@greenweb.org
t: @mrchrisadams

German Office
Naunynstrasse 40
10999 Berlin
Germany

See our contact page for more details
https://www.thegreenwebfoundation.org/contact/

Book a short call with me to discuss something.
https://cal.com/mrchrisadams