As part of its Global Logistics Provider activities, Daher is to drive the storage space for ITER. In this framework Daher manages the routing of parts from member countries and their storage before installation.
The Industrial Project is complex because this activity is marked by the low recurrence of incoming flows and the difficulty of anticipating their evolutions.
The old planning process had become ineffective:
Hypotheses on incoming flows were based on incomplete and complex data to be reconciled and manually integrated
Multiple data sources with different data formats
Use of a simulation tool that wasn't not very flexible, that didn't allow to integrate new constraints, with a long parameterization, and that didn't allow for multi-scripting
Lack of visibility of incoming / outgoing flows
The impact on storage capacity is very important.
DCbrain has set up a process that simply simulates the filling and scheduling of the site with the implementation of a Reinforcement Learning algorithms to calculate and optimize the different storage spaces according to the variable size of the items, the operators and the re-calculation according to the entrants.
The results of this modeling are as follows:
Forecast of the filling rate of the different storage spaces
Ability to change input data or add a new constraint
Scenario proposal (via the reinforced learning)
Download our white paper on supply chain optimisation thanks to Artificial Intelligence: