Extraction of key parameters and simplification of sub-system energy models using sensitivity analysis in subway stations

Abstract

Energy consumption of subway stations has become a focus of scientific research. For energy modeling, complete coverage of inputs makes it difficult to collect information for large-scale engineering application. To this consideration, this study conducted sensitivity analysis (SA) to determine key inputs and simplify energy model of subway stations. Firstly, four commonly-used SA methods (PEAR, SPEA, Morris and Sobol) were compared. Secondly, according to sensitivity indices, key inputs were extracted for ventilation and air-conditioning (VAC), lighting (LIGHT) and vertical transport (TRANS) model. For VAC model, the most important factors are humidity and temperature of the outside air and mechanical fresh air volume, contributing 40%, 33% and 11% of the change on energy simulation results, respectively. For LIGHT model, lighting power density of the hall exerts the greatest impact, with 50% of the change. For TRANS model, key parameters include train departure density and number of passengers, contributing 45% and 32% of the change, respectively. These outcomes provide guidance for engineers in the field of energy-saving operation for subway stations. Eventually, simplified VAC and TRANS energy models were investigated. Results indicate that CV-RMSE for simplified models are within 7.6%. The simplification greatly reduces the workload of data collection and model simulation.