A Random Forest Regression Predictive Approach for Determining the Mechanical Properties of Powders Using Microindentation: A First Step towards Improved Cold Spray Additive Manufacturing Quality Control

Product information

PhD Candidate Parisa Hasanpour Dastjerdi, phasa011@uottawa.ca, University of Ottawa; Prof. Hamid Jahed Motlagh, hamid.jahed@uwaterloo.ca, University of Waterloo; Postdoctoral Fellow Aleksandra Nasdic, anast064@uottawa.ca, University of Ottawa; Master Student Yeongmin Pyo, yeongmin.pyo@uOttawa.ca,University of Ottawa; Bertrand Jodoin, Bertrand.Jodoin@uottawa.ca, University of Ottawa ; Jing Hu; Verena Kantere; Luc Pouliot; Julia Villafuerte; Murray Pearson 

https://doi.org/10.53192/ITSC2026516

Powdered materials used in Cold Spray Additive Manufacturing (CSAM) exhibit mechanical behaviors distinct from their bulk counterparts, yet bulk properties are repeatedly assumed due to the lack of accessible powder-specific data. This leads to inaccurate simulations and deviations in CSAM product performance. While experimental methods such as single-particle compression and micro-tensile testing exist, they are complex, time-consuming, and largely limited to academic research. To overcome this limitation and support practical, daily use by CSAM operators, we propose a predictive approach that combines validated finite element (FE) simulations with a Random Forest Regression AI model.  The model is trained on FE-generated data and estimates key constitutive parameters of powders with the sole input of powder microindentation dimensions. This method is a first step to offer a fast, reliable tool for assessing powder quality, enabling users to detect batch-to-batch variability and avoid costly spray trials with substandard powders.

Reihe

PDF-Download zu https://doi.org/10.53192/ITSC2026516

Erscheinungsdatum

March 2026

Bindung

PDF Download