Topology Optimization for Transient Wave Propagation Problems

The study of elastic and optical waves together with intensive material research has revolutionized everyday as well as cutting edge technology in very tangible ways within the last century. Therefore it is important to continue the investigative work towards improving existing as well as innovate new technology, by designing new materials and their layout.

The thesis presents a general framework for applying topology optimization in the design of material layouts for transient wave propagation problems. In contrast to the high level of modeling in the frequency domain, time domain topology optimization is still in its infancy. A generic optimization problem is formulated with an objective function that can be field, velocity, and acceleration dependent, as well as it can accommodate the dependency of filtered signals essential in signal shape optimization [P3].

The analytical design gradients are derived by use of the adjoint variable method. Many wave propagation problems are open-region problems, i.e. the outer boundaries of the modeling domain must be re ection-less. The thesis contains new and independent developments within perfectly matched layer techniques for scalar as well as for vectorial elastic wave propagation problems using finite element analysis [P2], [P4].

The concept is implemented in a parallel computing code that includes efficient techniques for performing gradient based topology optimization. Using the developed computational framework the thesis considers four optimization problems from nano-photonics: First, an optical taper [P1] and a notch filter [P2] - both optimized by energy maximization.

The last two cases demonstrate pulse shaping and delay in one [P3] and two [P5] dimensions. Whereas the test problem in [P3] is rather academic, the example considered in [P5] optimizes structures that accommodate non-dispersive slow light, with important applications for optical buffering devices.