Lawrence Berkeley National Laboratory

In recent decades, environmental problems have enforced designers to estimate the level of environmental emission of building design and reduce their environmental impact. On the premise of ensuring indoor comfort, the cost-effectiveness of solutions for reducing the building's greenhouse gas has become a critical issue. Based on the Life Cycle Assessment (LCA), this paper establishes a building performance trade-off framework for indoor thermal comfort, economics, and environmental implication. This framework consists of four parts: the establishment of the optimization model; sensitivity analysis; obtain of Pareto frontier solutions, and decision-making analysis. Optimization variables involve envelope type and some envelope physical parameters. The “design variables-building performances” database is obtained by using building simulation software combined with the Latin hypercube sampling algorithm. Sensitivity analysis is used to extract the key factors affecting building performance. The designer can prioritize these key factors and it can reduce the uncertainty of building performance. A multi-objective optimization method coupling Gradient Boosted Decision Tree (GBDT) and non-dominated sorting genetic (NSGA-II) algorithm is proposed to seek the trade-off between three performances (obtain Pareto frontier solutions). The Pareto solution provides a more comprehensive reference for the preferences of different stakeholders, and the set of alternative solutions is further shrunk. Finally, take a specific residential building in China's cold climate zone as a showcase of the trade-off framework. According to the obtained Pareto frontier solution, the solution set is shrunk to a certain range, and the distribution ranges of Life Cycle Costs, the greenhouse gas emissions, and the annual thermal discomfort hour ratio are 122.3–137.1 USD/m2, 15.6–44.8 kg CO2/m2, and 19.1–25.2%, respectively. The trade-off framework adopts the order of objective Pareto optimal and then subjective preference selection, narrowing the scope of alternatives for designers and saving time-cost of decision-making.