Analysis of surimi extrusion behavior during 3D printing by modified Computational Fluid Dynamics (CFD) and quick prediction of printability using machine learning based on texture data

Abstract

The purpose of this study was to develop a modifying Computational Fluid Dynamics (CFD) model for precise analysis of surimi extrusion behavior and establish a machine learning method for quickly prediction of surimi printability. The 2D rotational axisymmetric model used in this study to replace the 3D/4D physical model, which could improve the simulation efficiency. To improve the accuracy of the CFD simulations, the wall “no-slip” boundary condition assumed in the textbook was replaced with a boundary condition of wall slip. This reduced the error between the simulated results and the measured ones to <1%, and thus calculated required extrusion pressure below 17,034 Pa for printability. The CFD calculation efficiency was improved 33 times than the previously reports by the modifying CFD model, and the simulated results for the required extrusion pressure of inks could be obtained within 10 s. Additionally, a machine-learning method (partial least squares regression model) based on texture data was proposed to quickly predict the required extrusion pressure of surimi to assessed printability. The machine learning model showed a good performance with an R2 > 0.95.