An experimental analysis of deepest bottom-left-fill packing methods for additive manufacturing. Issue 22 (16th November 2020)

• Record Type: Journal Article
• Authors: Araújo, Luiz J.P.; Panesar, Ajit; Özcan, Ender; Atkin, Jason; Baumers, Martin; Ashcroft, Ian
• Is Part Of: International journal of production research. Volume 58:Issue 22(2020)
• Page Start: 6917
• Page End: 6933
• Publication Date: 2020-11-16
• Subjects: Build volume packing -- benchmarking -- additive manufacturing -- 3D printing -- deepest bottom-left-fill; Factory management -- Periodicals; 658.57
• DOI: 10.1080/00207543.2019.1686187
• Languages: English
• ISSNs: 0020-7543

Abstract

Abstract : The adoption of Additive Manufacturing (AM) technology requires the efficient utilisation of the avail- able build volumes to minimise production times and costs. Three-dimensional algorithms, particularly the Deepest Bottom-Left-Fill (DBLF) heuristic, have been extensively used to tackle the problem of packing arbitrary 3D geometries within the AM sector. A particularly common method applied to more realistic packing problems is the combination of DBLF and metaheuristics such as Genetic Algorithms (GAs). Through a series of experiments, this paper experimentally investigates the practical aspects, and comparative performance of different DBLF based methods including a brute force algorithm and GA combined with DBLF for AM build volume packing. The insights into the relationship between algorithm efficiency (in terms of volume utilisation), simulation runtime, and practical requirements, in particular geometry rotation constraints are investigated. In addition to providing an increased comprehension of the practical aspects of applying DBLF algorithms in the AM context, this study confirms the limita- tions of traditional DBLF and the requirements for more flexible and intelligent placement strategies while experimentally demonstrating that higher degrees of freedom for part rotation contribute to small improvements in volume density. The resulting additional computational effort discourages this strategy, however.