Geometric Algorithms for Machine Based Optimisation of Acoustic Reflectors

One of the challenges in optimizing acoustic reflector design in a performing arts venue is the potential interference with other elements in the room, notably sightlines for audience members or stage lighting. Conventional machine based optimization of geometric shapes like an acoustic reflector randomly augment the shape of the reflector inside rectilinear Bounding Boxes. The space available for acoustic reflectors in a theatre or concert hall rarely comes in the shape of a six sided rectilinear box. A collection of algorithms has been developed to address this issue, creating what might be called Bounding Breps. To test the geometries created within these B-Breps, a key component of any machine based optimization is the Objective or Fitness Function that the algorithm is working towards. An Objective/Fitness Function has been developed that optimizes reflected sound to arrive on a surface from lateral directions, at points evenly distributed on that surface. The Bounding Breps and the Objective/Fitness Functions were applied using both Simulated Annealing and Genetic Algorithms. The latter was found more appropriate for acoustic reflector design.