Building Airflow Optimization using Genetic Algorithm

Building performance analysis allows the architect to understand the behavior of buildings such as thermal and ventilation performance criteria. Computational Fluid Dynamics (CFD) is used to predict indoor thermal environments and assess their response to specific internal / external conditions and are currently employed in almost all stages of architectural design, construction, and management. To conduct the evaluation and optimization of the building thermal environment during early design process, parametric analyses can be performed. Parametric studies involve solving the problem several times with different sets of parameter variables such that the solutions can be found. It is difficult to build a CFD simulation for every change in the value of a parameter. Thus, the design space is difficult to explore systematically and solutions can be overlooked.

This led me to investigate and focus on developing simulation tools for use during early architectural design process. To allow efficient exploration of design alternatives that maximize performance ventilation requirements, I co-developed a design decision support system that integrated an evolutionary algorithm to a simulation tool was unique and allows new discovery of building shapes to suit thermal comfort criteria. The objective of this research was to allow the designer to explore and visualize the design evolution and its form generation simultaneously examining the diverse instances of the state space in relation to specific goal requirements.

Figure below shows the decision support evolution model.

GA3 GA1

Figure above shows morph-process GUI which is used to select the design instance which can then be simulated for optimal solution.

This work used Genetic Algorithm (GA) as the evolution algorithm and CFD as the evaluation mechanism. Further, it used an iterative approach that allowed designs to be evaluated using CFD simulation automatically to maximize several thermal and ventilation criteria. This work was published in Automation in Construction journal, paper titled, “Decision Support and Design Evolution: Integrating Genetic Algorithms, CFD and Visualization.”