About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
Additive Manufacturing Modeling, Simulation and Machine Learning
|
Presentation Title |
Gaussian process regression modelling and texture control during hot deformation of additively manufactured maraging steels |
Author(s) |
Jubert Pasco, Clodualdo Aranas Jr., Thomas McCarthy |
On-Site Speaker (Planned) |
Jubert Pasco |
Abstract Scope |
This work systematically investigates the grain topology evolution and hot deformation response of additively manufactured 18Ni(300) maraging steels at a temperature range of 850 °C to 1050 °C and strain rate range of 0.01 s−1 to 1 s−1. The flow stress values under experimental deformation conditions were predicted using a Gaussian process regression (GPR) model with maximum likelihood parameter estimation approach and compared with an Artificial neural network (ANN) model with a 3-6-6-1 architecture. Results show that columnar prior austenite grains exhibited localized flow deformation behavior and preferential nucleation of recrystallized laths aligned along <111> // BD. The superior performance of the GPR model compared to the ANN model during cross-validation can be attributed to its effective balance in penalizing both model fit and complexity. A net-shape fabrication technique combined with a high temperature forming route to develop dense structures with controlled local texture is proposed. |
Proceedings Inclusion? |
Planned: |
Keywords |
Additive Manufacturing, Machine Learning, Iron and Steel |