1280. Optimized Reel Utilization in Make-to-Order Production by 3D Printing
Invited abstract in session MA-21: Packing with practical constraints, stream Cutting and packing (ESICUP).
Monday, 8:30-10:00Room: Esther Simpson 2.12
Authors (first author is the speaker)
| 1. | Ilayda Celenk
|
| Industrial Engineering and Innovation Sciences, Eindhoven University of Technology | |
| 2. | Willem van Jaarsveld
|
| Eindhoven University of Technology |
Abstract
We study a variant of bin packing problem that arises in 3D printing. In particular, this study focuses on 3D printing in a make-to-order production system, where a limited number of slots for fixed-weight pieces of stock materials, called reels, are considered to manufacture components with 3D printers. We assign components to reels for production as the component requests arrive one by one. The aim is to determine an assignment policy that minimizes the average loss on reels in terms of weight, which occurs when the remaining weight on a reel is too low to produce a component and has to be discarded. Due to the online nature of the problem, we formulate it as a Markov Decision Process. We propose an index policy based on analytical insights into the problem. Moreover, we propose to use that index policy as a starting point for applying deep reinforcement learning to the problem, and find that this yields policies that have close-to-optimal loss.
Keywords
- Cutting and Packing
- Programming, Stochastic
- Critical Decision Making
Status: accepted
Back to the list of papers