Browsing by Subject "Lattice"
Now showing 1 - 3 of 3
- Results Per Page
- Sort Options
Item Mechanical property variation in metal lattice struts(2018-05) Dressler, Amber Dawn; Seepersad, CarolynDirect metal laser sintered lattice structures offer favorable tradeoffs between strength and weight, which are of interest to designers. However, manufacturing defects present throughout the lattices create significant variability in mechanical properties and part performance. The goal of this research is to improve the understanding of how defects impact mechanical properties to enhance designers’ ability to design metal lattice structures reliably. Before analyzing full lattices, it is important to understand how design parameters impact individual lattice struts during manufacturing. To test individual lattice struts, tensile specimens with five struts interrupting the gauge section were manufactured in three strut diameters and two build orientations. Testing different strut diameters investigates how defects change with feature size when all features are less than 1 mm in diameter. The two build orientations orient struts parallel to the build platform (horizontal) and perpendicular to the build platform (vertical). The horizontal struts represent the worst-case scenario since the struts are completely unsupported, which often leads to excessively weak, rough, or nonexistent structures in DMLS. In an actual lattice, there would be a range of orientations from horizontal to vertical depending on the lattice orientation. In addition to the strut samples, solid tensile samples were manufactured to evaluate the bulk material properties and associated variability at a larger size scale, less impacted by small defects. Before tensile testing the samples, non-destructive analysis was conducted on each sample, including methods include visual inspections, geometric measurements, density measurements, and computed tomography scans. Then, each sample was tension tested and imaged throughout testing. Following testing, the fracture surfaces and testing data were analyzed to investigate failure trends in search of a robust design. The CT scan data showed that the cross-sectional area of the struts was smaller than desired and that the horizontal struts were rougher than the vertical struts. Due to the poor quality of horizontal struts, the strut samples exhibited more variation in ultimate strength than the solid samples. Since five independent cracks are required to completely fracture a strut sample, the strut samples had less variation in percent elongation than the solid samples.Item Reduced-order process-aware design of multiscale structures with architected materials(2020-08-12) Sharpe, Conner Terry; Seepersad, Carolyn; Crawford, Richard; Haberman, Michael; Aughenbaugh, JasonAdvances in additive manufacturing (AM) techniques have enabled the production of parts with ever increasing geometric complexity across multiple length scales. However, while expanding the space of possible designs is exciting from an engineering perspective, it also presents new challenges that the design community must address to fully take advantage of these capabilities. Analyzing and searching through potential designs in a multiscale setting can quickly become intractable. This work seeks to alleviate these concerns by decoupling the design process across hierarchical length scales and developing surrogate models that can link these scales in a computationally efficient manner. Reduced-order design representations are used at each length scale to render a more tractable search space. Surrogate optimization combined with geometric projection approaches are used to yield architected materials (or unit cell geometries) with desirable homogenized mechanical properties. The macroscale material density field is parameterized using B-Spline surfaces to design lattice structures for stiffness and strength. Many of the most popular AM processes also produce inherently anisotropic properties that depend on geometric design parameters such as feature orientation or size. This factor is often ignored in the design process and consequently causes deviations in manufactured performance from desired outcomes, thus limiting the adoption of AM. This work builds a data-driven method to incorporate design-dependent properties produced by particular AM processes into a multiscale design optimization workflow. A study of the mechanical properties of struts built at various orientations and sizes in metal powder bed fusion is used to build a process-aware material model. Accounting for these manufacturing concerns through design is shown to result in significant improvements in the effective properties of architected materials and lattice structures. Finally, AM allows for an unprecedented freedom in the topology or material layout within a given design domain. However, it is often difficult for engineers to utilize this freedom by fully considering design possibilities as there is a dearth of tools that allow for efficient topology exploration. This work explores the use of deep learning algorithms as generative design tools, allowing for quick visual exploration of design options in the early stages of design. Specifically, conditional generative adversarial networks are used to encode a diverse suite of complex topologies derived from structural optimization in a compact latent space to enable many design options to be considered for the problem at hand. Design factors such as type of architected material, allowable mass, and loading conditions can be included in the exploration. These tools can also be used as reduced-order design parameterizations in themselves to enable further directed design when traditional optimization approaches may be difficult to implement effectively.Item Rheology and fabric in the continental lithospheric mantle from naturally deformed peridotites(2018-10-09) Bernard, Rachel Eleanor; Behr, Whitney M.; Barnes, Jaime D; Dygert, Nicholas J; Grand, Stephen P; Smith, DouglasLower crust and upper mantle xenoliths, which come to the surface through volcanic eruptions, give us a direct window into deep lithospheric deformation. By using techniques such as electron backscatter diffraction (EBSD), we can use xenoliths to place natural con- straints on the extrapolation of laboratory flow laws to Earth and inform geodynamic and geophysical models that rely heavily on knowledge of rheological structure. In Chapter 2, we use xenoliths from the tectonically active Mojave region of southern California to inform us of the nature of the lithosphere in this complex region. The results gleaned from xenoliths demonstrate that the mantle lid beneath the eastern Mojave is deforming by dislocation creep at relatively low stresses and fast strain rates, with an associated average viscosity of ~10¹⁹ Pa*s. The xenoliths also record a wide range of microstructures indicative of substantial strain localization at Moho depths. Finally, a study of the crystallographic or lattice preferred orientation (LPO) of crustal and mantle xenoliths reveal (1) that the mantle lid preserves both modern and fossilized anisotropy, and (2) crustal and mantle seismic fast axes may be anticorrelated across the Moho, even when both layers are kinematically coupled. In Chapter 3, we use these Mojave xenoliths in addition to several peridotites (a total of 65 samples) from a variety of geologic settings to study the broader relationship between olivine LPO and deformation conditions. Of particular interest are water content and stress, which past experiments have found strongly influence olivine LPO. Data from these samples, combined with an extensive literature compilation, reveal the extremely complicated nature of olivine LPO types and their link to deformation conditions. The results suggest that factors such as deformation geometry and history play a greater role in determining olivine LPO type than stress magnitude, temperature, and water content -- factors thought to be most influential based on experiments. In Chapter 4, we utilize the dataset of Chapter 3 to study the effect of orthopyroxene on bulk anisotropy. Among those samples with both olivine and orthopyroxene LPO, seven types of orthopyroxene LPO are identified. For each sample, measures of anisotropy are calculated from LPO data weighted for a range of hypothetical olivine and orthopyroxene modal percentages. Results reveal that for typical mantle peridotite, orthopyroxene LPO decreases anisotropy but does not affect the orientation of seismic fast directions.