Comparison and Integration of CFD-based Adjoint Method and Genetic Algorithm for the Inverse Design of Indoor Environment

摘要

Recently researchers have begun the combination of computational fluid dynamics (CFD) with optimization algorithms for indoor environment design. These optimization algorithms include genetic algorithm (GA) and adjoint method. It is necessary to quantitatively investigate the accuracy and efficacy of these design methods to identify their pros and cons. This study applied GA and adjoint methods for designing the air distribution in a two-dimensional (2D) ventilated cavity by using the air supply location and parameters as the design variables. The results show that the adjoint method was fast. The GA method was more accurate, but it required 20 times more computing time than the adjoint method. This investigation further integrated the adjoint and GA methods for designing thermal environment in an aircraft cabin. The GA method was used for initial design and the adjoint method for fine tuning. The integrated method converged easier than the GA method.