Ph. D. Project
METHOD - FroM the EarTH to the ClOuD : Decision support holistic approach for eco-designing smart systems
2022/10/27 - 2025/09/30
Other supervisor(s):
Limiting global warming to 2°C by 2100 compared to the pre-industrial era is the ambition of the Paris
agreements (COP21), with the roadmap to reduce greenhouse gas emissions from all human activities by
50% by 2030 compared to the reference year 2015 and to make electricity production cleaner with a
global emission rate of 200 gCO2e/Kwh by 2030. In this context, the digital sector has an ambivalent role,
as it enables the deployment of intelligent solutions to reduce the carbon footprint of cities, buildings,
transport, industry and also the energy sector to produce and transport electricity. However, this positive
contribution to the planet is counterbalanced by the fact that to create this intelligence, it is necessary to
rely on data centers, networks, connected objects, and other terminals that are costly in terms of energy
consumption and mainly electricity to power and cool this IT equipment. In 2015, global digital
consumption was about 1000TWh corresponding to the emission of 740 Mt of CO2e. For comparison,
France emitted 440 Mt of CO2e in 2019.

In addition to the energy aspects, there is the question of the criticality of materials. Indeed, even artificial
intelligence is based on objects that require mineral resources to manufacture. Awareness of the criticality
of materials is relatively recent and began in 2010 with the Chinese embargo on rare earth exports to
Japan. Since then, the European Commission has created a working group "Critical Raw Materials" to
establish a list of critical materials to guide industrial strategies, research and innovation. The current
shortage of semiconductors illustrates the importance of this topic. It should also be noted that the
extraction of these same materials is not neutral from an environmental point of view. Indeed, the
extraction of metals generates CO2 emissions and is a major consumer of water. While the massive use of
metal recycling, leading to a circular economy, will limit the environmental impact of their use, it can also
generate collateral damage in terms of by-products (e.g. Molybdenum and Rhenium for Copper, Indium for
Zinc), the extraction of which will be greatly limited. By combining geopolitical, environmental and
economic dimensions, the question of resources for the digital and energy transitions is as crucial as it is

Identification of the problem: CRAN's works on Green ICT have mainly focused on the usage phase of
digital solutions and essentially on their electricity consumption but are too reductive to establish a
complete environmental analysis. It is then necessary to consider methods to have a holistic view on the
impact of the digital solution but also on the constituents of the digital solution during its entire life cycle,
i.e. from the production phase to its dismantling. Standards exist to analyze this lifecycle (e.g., L-1410) but
they lack tools to:
(i) to draw "simple" "quantifiable" conclusions in a complex environment due to the multiplicity of criteria
to be considered ;
(ii) to be guided in the choice of technological and/or software components allowing the minimization of
the environmental impact of the system to be implemented, while maintaining the expected level of

There are already works to analyze the life cycle of systems based on multi-criteria methods, in particular
(as an example) to evaluate the sustainability of renewable energy technologies. However, the original
problematic that we seek to develop is to simultaneously integrate the costs and benefits of deploying
digital/smart solutions to make the considered application cleaner (industry, home, city, transportation...).
Objective of the thesis: The research work carried out in this thesis aims at designing a decision support
framework allowing intelligent system designers to evaluate and analyze the environmental balance of
their solution/architecture on the whole life cycle, but also to identify the optimal configuration(s), both
from a technological (physical) and software point of view.

Considering the whole life cycle starts with the extraction of resources from the earth to manufacture the
equipment of the envisaged solutions. It is then necessary to have an expertise of the domain to be able to
develop decision support tools when it comes to making choices. These choices will be guided by
environmental criteria including
(i) The efficiency of the materials used compared to another,
(ii) its life span,
(iii) its properties to be reconditioned or recycled,
(iv) the energy required for its extraction and transportation,
(v) its availability (land reserve),
(vi) its costs,
(vii) but also societal and geopolitical criteria.

A proof of concept of this material audit has been published by the IJL team in the journal Advanced
Functional Materials. In this article, a state of the art of the different technologies was established
according to the previously mentioned criteria but this type of audit would be improved by the use of
artificial intelligence techniques. Indeed, in order to develop solutions that will make the previously
mentioned sectors of activity cleaner, it is necessary to study and propose machine learning models for the
prediction of existing "material" resources and future market demands, in conjunction with an evaluation
that anticipates the use and dismantling phases.

Thus, the objective of the thesis work is to generalize this approach to other physical components (digital
or not), to extend the criteria to be considered and to integrate it in the complete life cycle of intelligent
systems which will allow to evaluate, to quantify the real contribution of the digital to create an
intelligence making the world cleaner.
Green ICT, Earth's resources, LCA, Intelligent systems
Modeling and Control of Industrial Systems