In this study, the compression behavior of the body-centered cubic with exterior and interior vertical struts (BCCZZ) lattice structure produced with Polylactic Acid (PLA) has been investigated using experimental, numerical, and machine-learning algorithms. When comparing digital image correlation and the ANSYS Static Structural numerical module, the measurements of deformation in the -Y direction taken from the top-right, top-left, middle-right, and middle-left points of the lattice structure are closely matched, with differences of 3.5%, 0.66%, 22.3%, and 12.69%, respectively. However, measurements from the bottom-left and bottom-right points show discrepancies of 49.17% and 58.91%, respectively. The lack of agreement between numerical and digital image correlation (DIC) analyses at the bottom-left and bottom-right points of the lattice structure is attributed to deformation in the lower section observed in the experimental study. The numerical study, modeling only elastic deformation, fails to account for broken regions' deformation adequately. Furthermore, the elastic deformation region has been comparatively investigated using experimental, numerical, and multilinear regression (MLR) models. Despite the MLR algorithm being trained with data from the compression test and achieving an R2 value of 0.97, numerical modeling is closer to the experimental results. Thus, for the first time in the literature, the compression behavior of the BCCZZ lattice structure made from PLA+ has been comparatively investigated using experimental, numerical, and machine learning methods.
Additive Manufacturing Lattice Structure Digital Image Correlation Numerical Modeling Machine Learning Compression Behavior.
In this study, the compression behavior of the body-centered cubic with exterior and interior vertical struts (BCCZZ) lattice structure produced with Polylactic Acid (PLA) has been investigated using experimental, numerical, and machine-learning algorithms. When comparing digital image correlation and the ANSYS Static Structural numerical module, the measurements of deformation in the -Y direction taken from the top-right, top-left, middle-right, and middle-left points of the lattice structure are closely matched, with differences of 3.5%, 0.66%, 22.3%, and 12.69%, respectively. However, measurements from the bottom-left and bottom-right points show discrepancies of 49.17% and 58.91%, respectively. The lack of agreement between numerical and digital image correlation (DIC) analyses at the bottom-left and bottom-right points of the lattice structure is attributed to deformation in the lower section observed in the experimental study. The numerical study, modeling only elastic deformation, fails to account for broken regions' deformation adequately. Furthermore, the elastic deformation region has been comparatively investigated using experimental, numerical, and multilinear regression (MLR) models. Despite the MLR algorithm being trained with data from the compression test and achieving an R2 value of 0.97, numerical modeling is closer to the experimental results. Thus, for the first time in the literature, the compression behavior of the BCCZZ lattice structure made from PLA+ has been comparatively investigated using experimental, numerical, and machine learning methods.
Additive Manufacturing Lattice Structure Digital Image Correlation Numerical Modeling Machine Learning Compression Behavior
Birincil Dil | İngilizce |
---|---|
Konular | Makine Mühendisliği (Diğer) |
Bölüm | Araştırma Makalesi |
Yazarlar | |
Yayımlanma Tarihi | 30 Nisan 2025 |
Gönderilme Tarihi | 27 Mayıs 2024 |
Kabul Tarihi | 12 Şubat 2025 |
Yayımlandığı Sayı | Yıl 2025 Cilt: 9 Sayı: 1 |
Uluslararası 3B Yazıcı Teknolojileri ve Dijital Endüstri Dergisi Creative Commons Atıf-GayriTicari 4.0 Uluslararası Lisansı ile lisanslanmıştır.