Multi-Material and Multi-Structural Optimisation for Generative Design of Controlled Property Domains

Yang, Wuxin
Singamneni, Sarat
Huang, Loulin
Item type
Degree name
Doctor of Philosophy
Journal Title
Journal ISSN
Volume Title
Auckland University of Technology

Most naturally evolved structures exhibit the highest level of optimisation in terms of the material properties through carefully selected and adopted compositional and structural variations resulting in the functional grading of the highest order in the materiality of the continuum. Engineered designs and products lack such finer adjustments in the material attributes, mainly due to the limitations of the manufacturing processes, which have by far been suitable for bulk material consolidation mostly. With the advent of additive manufacturing technologies, the point-by-point material consolidation realm has become a reality, giving renewed opportunities to revisit the functional material grading regimes. Considering the immense benefits of being able to design and manufacture products with varying material characteristics to suit to specific performance attributes, the multi-material design and manufacturing opportunities have attracted significant research attention in recent years. In particular, the multi-material printing based on the multi-jetting UV cured polymer processing theoretically allows for close control of the material composition and properties within a given printed part domain, allowing to design the material rather than designing a part with a given material for the target functionality. Based on these unique qualities, multi-material printing has been evaluated significantly, exploring the generative design dream to fabricate parts with carefully controlled material properties. However, the digitally mixed acrylic resins being inferior in mechanical properties, these promising developments were restrained to theoretical studies in most cases. The point-by-point or line-by-line material consolidation mechanics typical of all additive technologies can be further exploited to derive a step-change in the pursuit of generative material design to achieve tailored performance attributes. The current research addresses this gap by extending the multi-material dispersion problem to optimise the structural forms of voxelised geometries developed by discretising the problem domains of the design task. The challenge is still a generative design task and needs the use of multi-objective evolutionary algorithms integrated with the numerical evaluation schemes to explore the multi-material or multi-structural continuum mechanics. Beginning with the standard genetic algorithm, a variety of other evolutionary algorithms are used for the optimisation schemes, while the numerical simulations of the multi-material or multi-structural domains are implemented in COMSOL Multiphysics. The initial focus was on establishing the relative merits of different evolutionary algorithms development of the MATLAB coding that integrates the iterative interactions and evolutionary progress based on the optimisation and numerical simulation schemes. The multi-material case was solved during this stage, with a greater focus on the computational algorithms and processes. Once this was achieved, the structural manipulation of voxelised domains was undertaken for single material printing of generatively designed structural shapes using more advanced optimisation algorithms. The outcomes of the research are positive as an effective computational workflow was developed that will allow to generatively design a structural form with either multi-material or multi-structural optimised topologies for pre-set frequency, deflection, or other mechanical property responses.

Publisher's version
Rights statement