טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
Ph.D Thesis
Ph.D StudentMass Yoram
SubjectTopology Optimization Methods Adapted to Additive
Manufacturing Processes
DepartmentDepartment of Civil and Environmental Engineering
Supervisor Professor Oded Amir
Full Thesis textFull thesis text - English Version


Abstract

Topology Optimization (TO) deals with optimal repartition of material in a defined domain under given constraints. The algorithm outputs cannot usually be manufactured by conventional production means. Additive Manufacturing (AM) technologies allows a quasi free-form fabrication. One of the main restrictions of the current technologies is known as the “maximum overhang angle”, which is a geometrical limit above which supports are needed to form overhang patterns. Implementing AM constraints into TO algorithms such that the supports are part of the design is the most straightforward idea to marry them. Furthermore, because the printing direction affects directly the amount of supports, it must be considered in the optimization framework as well.

    In recent years, numerous methods promoting this support insertion concept were developed, with a few main approaches. The filter and the projection techniques condition existence of the material to the presence of its support in its immediate neighborhood. The boundary approach enforces the geometrical limitation through direct action on the layout boundaries, as they define the limit of the part hence if a pattern is overhanged. Another method predefines movable components - or voids - that arrange to solve the optimization problem, while geometrical constraints can be applied. Only recent publications focus on the physical processes that are the root causes of the manufacturing limitations.

    The research thesis discussed herein presents the three approaches that were developed to address the geometrical printing difficulties or the physical phenomena explaining them. The associated tools that were implemented are also discussed, especially the ones measuring performance and printability. In the first method, the geometric constraints are imposed on a discrete structure as a first step: the optimal truss solving the optimization problem contains only printable bars. This truss serves as a virtual skeleton that influences the continuum formulation of the problem to privilege material repartition at the bars’ locations. As the target is to develop an industry-ready algorithm, the procedure considers not only box design domains but also convex and non-convex ones. Furthermore, one of the main features of the method is to search for the optimal printing direction, which is considerably more computationally onerous in 3D complex design domains considering the increasing number of truss elements, and the need to explore the whole solid angle. Regarding the continuum optimization, advantage is taken from the use of large-scale parallel topology optimization capabilities. The influence and difficulties related to each of the mentioned parameters, their implementation and tuning are discussed likewise. In the second method, artificial self-weight is added to the formulation to force support creation. Good trade-offs between printability and performance are reported, and extension of the method to 3-D is currently under development. This procedure mimics the physical process of fused deposition modeling techniques, as in plastic printing. In the last approach, the printing bed technology, used for metal construction, is simulated in a simplified manner through an integration of the thermal compliance in the formulation of the optimization problem.