Paper Conference

Proceedings of Building Simulation 2017: 15th Conference of IBPSA


An Integrative Algorithmic Platform Coupled with Gradient Descent and Parametric Analysis Methods to Optimize Skylight Sizes

Sara Motamedi1,2, Petra Liedl1
1University of Texas at Austin
2Interface Engineering Inc, San Francisco

Abstract: While daylighting strategies show 20-70% savings in lighting loads, recent research has attempted to optimize fenestration by parametric analysis or Genetic Algorithm. Since both techniques need descent amount of iterations, they are time consuming and computationally expensive. In this paper we applied an optimization method of gradient descent to an integrative platform, coupling daylight and energy tools. We calibrated skylight sizes while minimizing total energy consumption. The results show that the gradient descent method converges faster with more accurate results. In addition, parametric analysis was applied to the integrative algorithm to optimize skylight sizes based on both daylight and energy performances.
Pages: 1956 - 1965