Harpe, Partitioning Models to Minimize the Parallel Print Time in Fused Filament Fabrication

Abstract

3D printers are used for various applications such as education, manufacturing and prototyping. Prototyping is a process where it is very useful to have a physical representation of the object that is being designed. A common workflow in prototyping is an iterative process where a design is printed. After the design has been printed flaws and possible improvements can be discovered. The design is adjusted and the process starts again. In this process a fast turnaround time is valued. For some models the printing process takes hours or even days. This slow turnaround time stagnates the iterative designing process.

Using multiple 3D printers could solve this issue. The model can be split into a number of parts that is at most equal to the number of available printers. All sub-models can then be printed in parallel. In order for this method to be effective, the model-splitting needs to be automated. Doing this manually would require additional time, defeating the purpose of the high turnaround time. Additionally the assembly process should be as easy as possible. Having a partitioning that requires an intensive assembly process can cost a significant amount of time that exceeds the time gained by the partitioning.

As all parts are printed simultaneously the total print time of all parts is determined by the part with the longest print time. The goal of the algorithm thus becomes an optimization problem of finding the partition of the input model containing at most n pieces that minimizes the print time of the slowest printed sub-model.

Contributions of this work include - a novel method to estimate the print time, while less precise than previous methods, it can predict the print times of models significantly faster, -a partition algorithm, that cuts a model in n parts, -a method calculates properties for a set of candidate cuts, making it possible to evaluate a dense collection of cuts while maintaining an efficient algorithm, -a local search procedure that improves an existing solution by iteratively optimizing the partitioning, and -a method for adding connectors between the model-parts for an increased ease of assembly.