A machine learning method to predict printing time for the L-PBF process - proposes Random Forest Regressor

Aug 3, 2026

A machine learning approach is presented for the Laser Powder Bed Fusion (L-PBF) process of metal powders, aiming to identify optimal parameter combinations and overcome the barrier of long printing times for complex parts. Early prediction of 3D printing time supports technical and economic component assessment, improving cost efficiency, production capacity, energy usage, environmental impacts, and lead time. The method uses Random Forest Regressor with STL CAD input and demonstrates high prediction accuracy through a case study.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages6

Reading Time15 min

AccessFree to download

Introduction

Additive manufacturing and L-PBF

Design for Additive Manufacturing (DfAM)

State-of-the-art

L-PBF process

[Available online at www.sciencedirect.com](Available online at www.sciencedirect.com) ScienceDirect Procedia CIRP 136 (2025) 671–676 35th CIRP Design 2025 A machine learning method to predict printing time for the L-PBF process Michele Trovatoa *, Michele Amicarellia, Daniele Ferraraa, Mariorosario Pristb, Paolo Cicconia a Università degli Studi Roma Tre, via della Vasca Navale 79, 00146 Rome, Italy b Università politecnica delle Marche, Via Brecce Bianche, 12, 60121 Ancona, Italy * [Corresponding author.](Corresponding author. E-mail address: michele.trovato@uniroma3.it)[ E-mail address:](Corresponding author. E-mail address: michele.trovato@uniroma3.it)[ michele.trovato@uniroma3.it](Corresponding author. E-mail address: michele.trovato@uniroma3.it) Abstract The machine learning usage in the L-PBF process for metal powders helps to identify the optimal parameter combination. Machine learning can find non-linear correlations between the high number of variables ofthis production process. One of the obstacles to the widespread adoption of L-PBF in the industry, in addition to the high costs, is the long printing time required for a complex component. The possibility of an early evaluation of the 3D printing time could promote the overall diffusion of this production process in the industry. Correct time prediction can improve cost efficiency and production capacity, reducing energy consumption, environmental impacts, and lead time. This paper proposes a machine learning approach, such as Random Forest Regressor, to predict the printing time of a metal component starting from the STereo Lithography (.stl) CAD format for the L-PBF process. A case study is proposed to evaluate and demonstrate the approach, obtaining a high-level prediction accuracy. © 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ([https://creativecommons.org/licenses/by-nc-nd/4.0](https://creativecommons.org/licenses/by-nc-nd/4.0)) Peer-review under responsibility of the scientific committee of the 35th CIRP Design 2025 Keywords: Printing Time; Machine Learning; L-PBF Process; Design for Additive Manufacturing. 1. Introduction Additive Manufacturing (AM) is a set of production techniques characterized by being realized layer by layer. Standard ISO/ASTM 52900:2022 defines seven process categories: Binder Jetting, Direct Energy Deposition, Material Extrusion, Material Jetting, Powder Bed Fusion, Sheet Lamination, and Vat Photopolymerization [1] . AM enables the realization of complex and free-form geometries that are difficult or impossible to realize with traditional manufacturing techniques. Materials involved are polymers, ceramics, composites, pure metals, and alloys [2] . The high costs and long printing time, especially for the metal sector, are limiting the diffusion of AM in the industry. However, some factors are overturning this trend such as the high tooling costs related to subtracting manufacturing, the possibility of ongoing changes during the project phases, shorter time to market (agile reaction in all steps of the product life cycle), the design complexity allowed, the possibility of low/medium volumes production, the digital inventory, and the possible product customization. Design for Additive Manufacturing (DfAM) is the discipline that studies tools and methods to overcome all technological constraints, related to materials and processes, and to optimize geometries [3] . AM is a digital process because it is based on the integration of digital tools, software, and workflows throughout the production cycle. Every phase of production can be related toa digital environment; the design phase uses CAD software to realize a 3D digital model that can be optimized using algorithms; the slicing phase and process planning such as the nesting phase, material deposition, layering, etc. are entirely digital and associated to the printer software, etc. [4] . The p

FAQs

Why is predicting printing time important for the L-PBF process?

Long printing time limits the industrial adoption of L-PBF. Accurate early prediction improves cost efficiency, production capacity, energy consumption, environmental impacts, and lead time.

What inputs does the proposed model use to predict printing time?

The approach uses STL CAD format as the input for the L-PBF process, enabling time prediction from geometry-related information.

Which machine learning method is proposed in the paper?

The paper proposes a Random Forest Regressor-based machine learning method to evaluate and predict the printing time for metal components in L-PBF.

Aug 3, 2026

Upload by Valentina