POMPIER Research Project

Planning, Scheduling and Maintenance 
for Integrated and Robust Production

Group : Reliability Engineering and Decision-Making tools

Labelling: none

Duration: (2023 - 2027)

Funding: ANR

Staff involved from LARIS: Bruno CASTANIER (project manager)

Project partners:

Abstract:

Many industrial players are convinced of the substantial benefits to be gained by converging the optimization objectives associated with production and maintenance processes. In spite of this, traditional production management approaches, which serve as a reference for definishing and steering these processes at the tactical level, persist in treating these two concerns independently. What's more, these approaches are generally based on an aggregated vision of dynamic phenomena and operational decisions, which is at odds with the new paradigms of responsiveness and flexibility that govern industrial processes.

To address this issue, we propose to develop decision-support methods for jointly optimizing production and maintenance policies, taking into account the uncertainties inherent in manufacturing systems in order to design tactical plans that are both economically efficient and robust in the face of uncertainties. It is this desire for integration that motivates our proposal, driven by the following two observations:

1. The highly aggregated estimation of production capacities in the planification process too often leads to the definition of unattainable production targets, which have to be modified a posteriori according to operational data;
2. Operational agility, linked to real-time data acquisition, for scheduling and maintenance does not, paradoxically, profite tactical planification.

The general research hypothesis is that the integration of prescription rules derived from maintenance into tactical and operational decision models makes it possible to solve the joint problem of planification-ordering-maintenance. The general approach will be to introduce dependencies between operational efficacity and the health status of the production system to the classical aggregations of information usually taken into account by tactical planification. More specifically, we will focus on analyzing the random characters inherent in manufacturing systems and the modeling uncertainties linked to the level of knowledge of the phenomena that govern them.

The targeted scientific challenges are presented in the form of 4 research questions:

1. How to characterize, model and evaluate maintenance-derived prescriptive rules for scheduling and their potential for resolving the integration of different decision-making levels?
2. How to manage time scales inherent in planification horizons through the use of a "digital shadow"?
3. How can we model the dependencies between scheduling efficacity and the health status of the production system afin to implement conditional and predictive maintenance policies?
4. How to characterize, model and assess the robustness of a scheduling solution in the face of uncertainties of a random and epistemic nature?