From simulation to data-driven approach: A framework of integrating urban morphology to low-energy urban design

Energy-efficient urban design is an important prerequisite to sustainable urban development and reduction of greenhouse gas emissions. This study proposes an automatic framework to optimize urban design through the use of an urban building energy model. Three optimization goals were defined: maximum solar energy utilization, solar lighting of the first floor, and minimum building energy demand. Urban morphology was integrated into the optimization process as the bridge between the urban design scenario and the actual urban block. To validate the model, this study abstracted basic urban forms from actual urban contexts to generate urban blocks with the Rhino tool and run optimization in the Wallacei X, for multi-objective optimization in Rhino. The long short-term memory (LSTM) network was applied to infer energy performance of 41 actual urban blocks in Jianhu, China. In the results, the proposed framework can be validated feasibly with optimization of 100 iterations. A set of optimal results will be achieved for three goals and five clusters defined for different concerns of urban design strategies. The LSTM can achieve the best accuracy of 1.21% and 1.37% for energy generation of photovoltaic and total building energy use intensity respectively.