Preform design to reduce forging load and grain size simultaneously in disk forging of IN718

Abstract

In this study, we propose a new methodology for designing preform shapes for turbofan blisk and disk forgings of IN718 superalloy using a CNN (Convolutional Neural Network). The objective function of the preform design model is set to minimize grain size and reduce forging load. Various preform shape candidates were generated using NURBS (Non-Uniform Rational B-Splines) curves, defined by points on the target shape outline, and the grain size was calculated through FE (Finite Element) analysis based on the JMAK (Johnson-Mehl-Avrami-Kolmogorov) model to train the CNN model. The preform shapes derived from the proposed load and grain size-aware CNN model were compared and validated against those from previous studies regarding grain size and load reduction. An objective function f, which is capable of quantitatively expressing the load and grain size improvement, was developed for comparison and validation. The results showed an improvement of 19.64% for the blisk and 18.96% for the disk compared to previous studies. This research is expected to significantly contribute to the forging of aerospace components.

Keywords

• IN718
• Disk forging
• Preform
• JMAK model
• CNN