熔模铸造
材料科学
陶瓷
建模与仿真
工艺优化
过程(计算)
工艺工程
机械工程
仿真建模
过程建模
铸造
冶金
投资(军事)
数值模拟
计算机模拟
制造工程
过程模拟
工程制图
多目标优化
复合材料
过程变量
合金
砂型铸造
数学模型
数据建模
多尺度建模
作者
Xiaoyu Hu,Jian He,Man Yuan,Qiongyuan Zhang,吴林,Haiyang Wang,Qihang Yang,Lixian Lian,Tao Jing,Qingyan Xu
标识
DOI:10.1016/j.jmrt.2026.07.268
摘要
Ceramic cores are used to form the internal cooling channels of hollow turbine blades, and their molding quality directly affects the dimensional accuracy of the final cast blade. However, ceramic injection molding is highly sensitive to processing parameters, making conventional trial-and-error optimization inefficient and costly. Although numerical simulation is widely used for process optimization, the polymer-oriented constitutive models embedded in commercial software may not adequately describe highly filled ceramic feedstocks. In this study, a customized Power Law–Cross–WLF viscosity model and a modified single-domain Tait PVT model were developed for a ceramic feedstock and implemented in Autodesk Moldflow Insight 2021 via the solver API. Short-shot experiments showed good agreement between the experimentally observed and simulated flow fronts, supporting the reliability of the customized simulation framework. A multi-objective optimization strategy integrating a Box–Behnken design, response surface modeling, an improved particle swarm optimization, and CRITIC–TOPSIS decision making was developed to simultaneously optimize volumetric shrinkage, warpage deformation, and sink marks estimate. Within the investigated processing window, melt temperature and injection rate had relatively limited effects, whereas packing pressure and packing time were the dominant factors. The recommended packing pressure and packing time were 1.61 MPa and 15 s, respectively, and the average agreement between the surrogate-model predictions and the subsequent Moldflow results was 95.65%. Relative to the initial processing condition, the optimized parameters reduced warpage deformation, volumetric shrinkage, and sink marks estimate by 38.42%, 12.05%, and 37.50%, respectively.
科研通智能强力驱动
Strongly Powered by AbleSci AI