Data-driven modeling for the correlation of the inputs and outputs in thermoplastic micro injection molding

Alireza Mollaei Ardestani*, Reza Asadi, Uma Maheshwaran Radhakrishnan, Inigo Flores Ituarte, Murat Külahci, Matteo Calaon, Jesper Henri Hattel, Guido Tosello

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Abstract

This paper explores the application of micro manufacturing in the production of plastic parts, focusing on the widely used injection molding process. The increasing demand for high-quality parts in industrial settings has led to a heightened need for digital twins in micro injection molding. To address this demand, a Data-Driven approach is employed, involving the simulation of process parameters effects in plastic injection molding. The project employs the Design of Experiment (DOE) methodology for a specific geometry, varying three key input process parameters—Melt Temperature, Mold Temperature, and Injection Speed—across different material grades. Responses such as Part Weight, Cavity Injection Time, and Maximum Injection Pressure are simulated using a commercially available Finite Element Analysis (FEA) Simulation software. Data Driven Modelling is achieved by incorporating viscosity and pvT coefficients of each material, along with the specified process parameters. Statistical Analysis, Machine Learning, and Deep Learning methods are employed for the data driven modeling. The results indicate that Part Weight and Maximum Injection Pressure are influenced by all three input parameters, while Cavity Injection Time is primarily affected by the Injection Speed of the machine. Both Statistical and artificial intelligence models demonstrate effective performance with the selected materials. Importantly, these models successfully predict results for materials not initially considered, affirming the achievement of Data Driven Modelling for the specific geometry under investigation.
Original languageEnglish
Title of host publicationEuropean Society for Precision Engineering and Nanotechnology, Conference Proceedings : 24th International Conference and Exhibition, EUSPEN 2024
Number of pages4
Publishereuspen
Publication statusAccepted/In press - 2025
EventEuropean Society for Precision Engineering and Nanotechnology, Conference Proceedings: 24th International Conference and Exhibition, EUSPEN 2024 - Dublin, Ireland
Duration: 10 Jun 202414 Jun 2024

Conference

ConferenceEuropean Society for Precision Engineering and Nanotechnology, Conference Proceedings
Country/TerritoryIreland
CityDublin
Period10/06/202414/06/2024

Keywords

  • Plastic injection molding
  • Design of experiments
  • Machine learning
  • Digital twin
  • Process optimization

Fingerprint

Dive into the research topics of 'Data-driven modeling for the correlation of the inputs and outputs in thermoplastic micro injection molding'. Together they form a unique fingerprint.

Cite this