Face to the growing awareness of environmental concerns issued from human activities, eco-design aims at offering a satisfying answer in the products and services development field. However when the considered products become complex industrial systems, there is a lack of adapted methodologies and tools. These systems are among others characterised by a large number of components and subsystems, an extremely long and uncertain life cycle, or complex interactions with their geographical and industrial environment. This change of scale actually brings different constraints, as well in the evaluation of environmental impacts generated all along the system life cycle (data management and quality, detail level according to available resources...) as in the identification of adapted answers (management of multidisciplinary aspects and available resources, players training, inclusion in an upstream R&D context...). So this dissertation aims at developing a methodology to implement ecodesign of complex industrial systems. A general methodology is first proposed, based on a DMAIC process (Define, Measure, Analyse, Improve, Control). This methodology allows defining in a structured way the framework (objectives, resources, perimeter, phasing...) and rigorously supporting the ecodesign approach applied on the system. A first step of environmental evaluation based on Life-Cycle Assessment (LCA) is thus performed at a high systemic level. Given the complexity of the system life cycle as well as the exploitation variability that may exist from one site to another, a scenario-based approach is proposed to quickly consider the space of possible environmental impacts. Scenarios of exploitation are defined thanks to the SRI (Stanford Research Institute) matrix and they include numerous elements that are rarely considered in LCA, like preventive and corrective maintenance, subsystems upgrading or lifetime modulation according to the economic context. At the conclusion of this LCA the main impacting elements of the system life cycle are known and they permit to initiate the second step of the eco-design approach centred on environmental improvement. A multidisciplinary working group perform a creativity session centred on the eco-design strategy wheel (or Brezet wheel), a resource-efficient eco-innovation tool that requires only a basic environmental knowledge. Ideas generated during creativity are then analysed through three successive filters allowing: (1) to pre-select and to refine the best projects; (2) to build a R&D projects portfolio thanks to a multi-criteria approach assessing not only their environmental performance, but also their technical, economic and customers' value creation performance; (3) to control the portfolio balance according to the company strategy and the projects diversity (short/middle/long term aspect, systemic level...). All this work was applied and validated at Alstom Grid on electrical conversion substations used in the primary aluminium industry. The methodology deployment has allowed initiating a robust eco-design approach recognized by the company and finally generating a portfolio composed of 9 eco-innovative R&D projects that will be started in the coming months.

Face à l’émergence des problématiques environnementales issues des activités humaines, l’écoconception s’attache à offrir une réponse satisfaisante dans le domaine de la conception de produits et services. Cependant, lorsque les produits considérés deviennent des systèmes industriels complexes, caractérisés entre autres par un grand nombre de composants et sous-systèmes, un cycle de vie extrêmement long et incertain, ou des interactions complexes avec leur environnement géographique et industriel, un manque évident de méthodologies et d’outils se fait ressentir. Ce changement d’échelle apporte en effet des contraintes différentes aussi bien dans l’évaluation des impacts environnementaux générés au cours du cycle de vie du système (gestion et qualité des données, niveau de détail de l’étude par rapport aux ressources disponibles…) que dans l’identification de réponses adaptées (gestion de la multidisciplinarité et des ressources disponibles, formation des acteurs, inclusion dans un contexte de R&D très amont…). Cette thèse vise donc à développer une méthodologie de mise en œuvre d’une démarche d’éco-conception de systèmes industriels complexes. Une méthodologie générale est tout d’abord proposée, basée sur un processus DMAIC (Define, Measure, Analyse, Improve, Control). Cette méthodologie permet de définir de manière formalisée le cadre de la démarche (objectifs, ressources, périmètre, phasage…) et d’accompagner rigoureusement l’approche d’écoconception sur le système considéré. Une première étape d’évaluation environnementale basée sur l’Analyse du Cycle de Vie (ACV) à haut niveau systémique est ainsi réalisée. Etant donnée la complexité du cycle de vie considéré et la variabilité d’exploitation d’un système industriel d’un site à l’autre, une approche par scénario est proposée afin d’appréhender rapidement l’étendue possible des impacts environnementaux. Les scénarios d’exploitation sont définis à l’aide de la matrice SRI (Stranford Research Institute) et intègrent de nombreux éléments rarement abordés en ACV, comme la maintenance préventive et corrective, la mise à niveau des sous-systèmes ou encore la modulation de la durée de vie du système en fonction du contexte économique. A l’issue de cette ACV les principaux postes impactants du cycle de vie du système sont connus et permettent d’entreprendre la seconde partie de la démarche d’éco-conception centrée sur l’amélioration environnementale. Un groupe de travail multidisciplinaire est réuni lors d’une séance de créativité centrée autour de la roue de la stratégie d’éco-conception (ou roue de Brezet), un outil d’éco-innovation peu consommateur de ressources et ne nécessitant qu’une faible expertise environnementale. Les idées générées en créativité sont alors traitées par trois filtres successifs, qui permettent : (1) de présélectionner les meilleurs projets et de les approfondir ; (2) de constituer un portefeuille de projets de R&D par une approche multicritère évaluant leur performance environnementale, mais également technique, économique et de création de valeurs pour les clients ; (3) de contrôler l’équilibre du portefeuille constitué en fonction de la stratégie de l’entreprise et de la diversité des projets considérés (aspects court/moyen/long terme, niveau systémique considéré…). L’ensemble des travaux a été appliqué et validé chez Alstom Grid sur des sous-stations de conversion électrique utilisées dans l’industrie de l’aluminium primaire. Le déploiement de la méthodologie a permis d’initier une démarche solide d’écoconception reconnue par l’entreprise et de générer au final un portefeuille de 9 projets de R&D écoinnovants qui seront mis en œuvre dans les prochains mois. / Face to the growing awareness of environmental concerns issued from human activities, eco-design aims at offering a satisfying answer in the products and services development field. However when the considered products become complex industrial systems, there is a lack of adapted methodologies and tools. These systems are among others characterised by a large number of components and subsystems, an extremely long and uncertain life cycle, or complex interactions with their geographical and industrial environment. This change of scale actually brings different constraints, as well in the evaluation of environmental impacts generated all along the system life cycle (data management and quality, detail level according to available resources…) as in the identification of adapted answers (management of multidisciplinary aspects and available resources, players training, inclusion in an upstream R&D context…). So this dissertation aims at developing a methodology to implement ecodesign of complex industrial systems. A general methodology is first proposed, based on a DMAIC process (Define, Measure, Analyse, Improve, Control). This methodology allows defining in a structured way the framework (objectives, resources, perimeter, phasing…) and rigorously supporting the ecodesign approach applied on the system. A first step of environmental evaluation based on Life-Cycle Assessment (LCA) is thus performed at a high systemic level. Given the complexity of the system life cycle as well as the exploitation variability that may exist from one site to another, a scenario-based approach is proposed to quickly consider the space of possible environmental impacts. Scenarios of exploitation are defined thanks to the SRI (Stanford Research Institute) matrix and they include numerous elements that are rarely considered in LCA, like preventive and corrective maintenance, subsystems upgrading or lifetime modulation according to the economic context. At the conclusion of this LCA the main impacting elements of the system life cycle are known and they permit to initiate the second step of the eco-design approach centred on environmental improvement. A multidisciplinary working group perform a creativity session centred on the eco-design strategy wheel (or Brezet wheel), a resource-efficient eco-innovation tool that requires only a basic environmental knowledge. Ideas generated during creativity are then analysed through three successive filters allowing: (1) to pre-select and to refine the best projects; (2) to build a R&D projects portfolio thanks to a multi-criteria approach assessing not only their environmental performance, but also their technical, economic and customers’ value creation performance; (3) to control the portfolio balance according to the company strategy and the projects diversity (short/middle/long term aspect, systemic level…). All this work was applied and validated at Alstom Grid on electrical conversion substations used in the primary aluminium industry. The methodology deployment has allowed initiating a robust eco-design approach recognized by the company and finally generating a portfolio composed of 9 eco-innovative R&D projects that will be started in the coming months.