Abstract

Building performance simulation can be integrated with optimization to achieve high-performance building design objectives such as low carbon emission and cost-effectiveness by holistically considering design variables across different disciplines. However, the complexity of the design problem increases greatly with increasing dimensionality. In some cases, solving high-dimension problems is not technically feasible nor time-efficient. Decomposition is one way to reduce the complexity and dimensionality of optimization problems. However, the decomposed optimization might achieve local optimum. Therefore, deploying appropriate decomposition strategies to achieve global optimum is paramount. This study investigates the deployment of hierarchical and parallel decomposition for building design optimization problems to ensure identification of global optimum. The feasibility of combining sensitivity analysis and decomposition is also explored. At the end of this study, some recommendations are given to help select an appropriate approach in practice.
First, this thesis proposes a hierarchical decomposition. Hierarchical decomposition divides an optimization problem into several interconnected subproblems solved sequentially. The proposed approach is applied to the multi-objective optimization problem that minimizes buildings' operating costs and carbon emissions. The results show that the hierarchical decomposition approach can reduce the number of simulations while achieving global optimums.
Second, this thesis proposes a parallel decomposition. Parallel decomposition divides the original problem into several smaller subproblems to be solved separately, and potentially, concurrently. The proposed parallel decomposition approach is applied to solve the single-objective optimization problems of a benchmark function and a low-rise office building. The results show that the proposed approach finds the global optimum and takes less computation time than optimization without decomposition.
Third, this thesis explores the feasibility of combining sensitivity analysis with decomposition for dimensionality reduction. The efficiency and accuracy of different methods are compared through three case studies.
The proposed hierarchical and parallel decomposition approaches can be applied individually or combined into a hybrid decomposition approach. This thesis concludes with some recommendations to help choose a decomposition approach to solve building design optimization problems.