Thesis Defense: May Thu Nwe Nwe, “Topology Optimization of Truss Structures Considering Stress and Stability Constraints”
THE DEPARTMENT OF CIVIL ENGINEERING
ADVISOR JAMIE GUEST, ASSOCIATE PROFESSOR
ANNOUNCE THE THESIS DEFENSE OF
May Thu Nwe Nwe
Thursday, August 1, 2019
“Topology Optimization of Truss Structures Considering Stress and Stability Constraints”
Topology optimization is a free-form approach to designing efficient structural layouts. Although highlighted repeatedly in literature for its ability to identify creative, high performance designs, it is also well known that oversimplification of the underlying optimization formulation can lead to impractical structural solutions. A common truss optimization problem formulation is to minimize linear elastic strain energy for a given structural mass (or minimize mass subject to a linear elastic deformation constraint). The optimal solutions obtained from such a formulation are independent of the direction of the applied load and often include members that may readily fail due to yielding or buckling, and/or colinear members with unbraced hinges that are unstable. Such solutions are, of course, impractical from structural engineering perspective. Incorporating stress and stability metrics into the topology optimization formulation, however, involves significant mathematical challenges primarily due to the possibility of vanishing members, which ultimately leads to singularities and disjointed regions in the design space that gradient-based optimization struggles to navigate. This thesis reviews and discusses these challenges and explores a disaggregated formulation where design and state variables are treated as free optimization variables to enable consideration of stress, (member) buckling, and global (system) stability metrics in a mathematically consistent manner. The impact of various constraints and objectives on the optimized solutions is demonstrated for simple truss design problems. The presented results clearly show that the incorporation of stress, member buckling, and global stability metrics results in more realistic design solutions. Future directions of research are also discussed.