Supply chains in the agri-food sector are highly globalized and often criticized for their lack of traceability, sustainability, and resilience. In response to these concerns, so-called “short” food supply chains are considered as an alternative. However, several studies highlight that the generally positive perception surrounding these chains is not based on solid scientific evidence. This doctoral project aims to provide scientific material to confirm or refute the hypothesis of the sustainability of short supply chains—and therefore their relevance as alternatives to traditional chains—as well as to propose a methodology, intended for decision-makers, to assess the real impact of a short supply chain on the three pillars of sustainable development.

The current literature, although relatively limited, has focused on the design, implementation, and coordination of the S&OP process. After more than 30 years, the S&OP process seems to have reached a turning point due to market changes, technological advances, and the proliferation of data. This shift is driven by the need for more effective risk management to create more resilient supply chains, especially in the face of frequent global crises (environmental, energy, geopolitical...) that have severely disrupted product and raw material flows. Short-term disruptions (demand peaks, raw material shortages) are expected to occur more frequently, with significant strategic impact.

The integration of S&OP with operational planning and execution is therefore crucial. In this context, many projects are underway to rethink forecasting and planning in supply chains, particularly emphasizing the exploitation of all available data to provide unprecedented predictive capabilities. This research examines the current state of S&OP and associated demand forecasting approaches. To prevent risks and improve decision-making, several studies have explored forecasting methods combining statistics and machine learning (AI). Other research has examined deep learning approaches to predict sales, or the use of data to improve forecasts in supply chains. Additionally, simulation models have been developed to identify risks and scenarios in logistics systems.

This project aims to develop forecasting models that mathematically integrate uncertainty. The goal is to identify the types of risks and opportunities specific to the automotive industry, capture relevant data, and combine them to minimize overall uncertainty. The thesis seeks to improve the quality of S&OP predictions by developing advanced models that incorporate uncertainty, identify industry-specific risks and opportunities, and propose effective strategies to reconcile commercial and industrial perspectives.

Prodeval is a leader in the production of biogas scrubbers, but its current logistics system (supply, production, distribution, installation), while efficient on a small scale, shows limitations in meeting growing demand and must triple its production by 2027, from 150 to 450 units per year.

Biogas scrubber production is complex due to the many stages required to ensure their quality and reliability, from design to assembly, including component manufacturing and performance testing. Each stage must meet strict safety and quality standards. Moreover, the dependence on a limited number of suppliers exposes the supply chain to risks in case of failure of a key actor. Currently, supply chain management does not guarantee resilient production capable of dealing with unexpected events and market fluctuations.

The production of biogas purifiers is complex due to the many stages required to ensure their quality and reliability, ranging from design to assembly, component manufacturing, and performance testing. Each stage must adhere to strict safety and quality standards. Additionally, reliance on a limited number of suppliers exposes the supply chain to risks in the event of a failure of a key player. Currently, supply chain management does not guarantee resilient production, meaning capable of facing unforeseen events and market fluctuations.

The first step will consist of a literature review on logistics configuration and management in the biogas industry. The second step will involve modeling Prodeval’s current supply chain to identify actors, processes, and potential disruption risks, based on field immersion. The third and final step will develop decision-support tools for resilient logistics management, using statistical methods or discrete-event simulation models with software such as Flexsim.

The thesis aims to (re)design a resilient supply chain for Prodeval, capable of managing risks and ensuring production continuity. Objectives include analyzing current vulnerabilities, defining resilience indicators, and developing decision-support methods to strengthen disruption management. The research will also propose innovative strategies to anticipate and manage logistical incidents, thus ensuring Prodeval’s long-term success in a fast-growing sector.

Savoye is a company providing hardware and software solutions for warehouses and logistics processes in general. Following its expansion into the US and Asian markets, it is currently experiencing strong growth.

The FLOWER project, coordinated by Savoye, involves the LISPEN (ENSAM Lille) and DISP (INSA Lyon) laboratories, aiming to integrate energy cost criteria into the design and algorithmic management of logistics platforms to significantly reduce CO2 emissions. Modern logistics platforms are organized around the “GTP” (Goods-To-Person) concept. In this system, order picking is performed by an operator at a dedicated station, continuously supplied with stock bins and order cartons. The scheduling of bins and order cartons, at the right time and in the right order, is ensured by high-speed Automated Storage and Retrieval Systems (ASRS), supported by patented proprietary algorithms. Current ASRS design and algorithm testing rely on commercial simulators, used to optimize management rules to maximize productivity. Thus, ASRS are sized to handle peak order flows but cannot adapt to downward variations or account for the environmental impact of operations. The climate crisis makes it imperative to include energy cost in real-time ASRS management. An increasing number of Savoye’s clients now include energy cost as a decision criterion in their calls for tenders.

The objective of this PhD is therefore to develop algorithms that ensure order preparation while reducing ASRS energy consumption. Two approaches are considered: reducing equipment speed and reducing travel distances. We will first study algorithms affecting speed to achieve significant energy savings while leaving most decision algorithms unchanged. Next, we will study distance-based algorithms to address decision problems such as:

If an item required by a picking station is available in multiple locations in the ASRS, from which location should its bin be retrieved to minimize overall energy consumption?

A Storage Location Assignment Problem (SLAP): in which physical location should the bin be restocked after picking?

This question involves not only finding the best way to organize ASRS stock for more than 100,000 references to reduce energy consumption, but also assessing the impact of this storage mission when the systems return to maximum performance. Particular attention will be paid to the applicability of the developed algorithms to real-sized instances.

This thesis aims to develop innovative indicators and optimization criteria to evaluate regeneration value chains, integrating the specificities of the plastic, textile, and battery sectors. The project is structured around: A literature review to inventory existing indicators and analyze the challenges of recycling. The proposal of new indicators and their evaluation through collaborative case studies. The development of multi-criteria frameworks to combine these indicators, in connection with the project's digital tools. Close collaboration with other working groups to ensure the coherence of methods.