Cognitive Building Systems. Optimization Algorithms in Architecture from Design to Production

TL;DR: The research proposes an integration of optimization apparatus called “ Cognitive Control System” (CCS) into a parametric design framework and examines several types of nonlinear solving algorithms, such as genetic algorithms, neural networks, and numerical mathematical solvers.

Abstract: Architects must deal with increasing amount of design constraints, which is the consequence of increasing demands on building’s performance in terms of sustainability and construction cost. On the other hand, complex geometries has become common part in architectural projects. Therefore, it is nowadays more true than before that the building’s qualities depend on architect’s ability to find the optimal solution for all, often contradicting constraints. This is a task for which due to the complexity necessitates the use of sophisticated solving algorithms integrated into the design workflow. The research proposes an integration of optimization apparatus called “ Cognitive Control System” (CCS) into a parametric design framework. Cognition or “ knowing ” is here defined in terms of the ability to respond to the performative criteria of a building by finding optimum solution. The CCS contains a set of global and local solvers. Its part is also an interface, the Interactive Graph Control (IGC) by which the user can steer and control the optimization process in a transparent fashion. This interactive platform presents the user not only the best optimal solution, but also the whole range of other possible solution, even if less optimal. The research examines several types of nonlinear solving algorithms, such as genetic algorithms, neural networks, and numerical mathematical solvers. The research reveals their pros and cons and demonstrates how these different types of algorithms can be integrated into parametric system to enhance the design process. The thesis presents how to set up an objective function for multiple objectives and how the function affects the optimization process. The functionality and usability of the solvers is demonstrated on several case studies. The case studies are performed on different scale projects with different solving complexity. The cases cover range of different geometrical and design topics, such as generating free-form roof structure with certain local height constraints, optimizing family house towards low energy consumption, daylight and cost or exploring the design options for museum building.