Reductive bottom-up urban energy computing supported by Multivariate Cluster Analysis

Abstract

(no german abstract available) The present research effort investigates the requirements of an urban energy computing environment, aimed to support strategic decision making with regard to physical and technological interventions as well as behavioral, and contextual changes. Providing an analytical overview of some previous efforts, the present contribution introduces a novel two-step approach towards bottom-up urban energy computing, involving a reductive phase and a re-diversification process. The reductive phase is performed through an automated process within a GIS platform. The developed process utilizes available large-scale data to generated an energy-relevant representation of the urban building stock. A matrix of energy-influential building characteristics, depicted as aggregate descriptive indicators, is computed based on the generated representation and subjected to Multivariate Cluster Analysis methods for stock classification. The resulting classes are represented through typical buildings, which undergo detailed performance simulation computations. The re-diversification process addresses the loss of diversity due to the reductive method, through employment of stochastic occupancy models and model parametrization. This paper reports on the 38 development of the reductive step, illustrating the encountered challenges and the adopted responses.

The present research effort investigates the requirements of an urban energy computing environment, aimed to support strategic decision making with regard to physical and technological interventions as well as behavioral, and contextual changes. Providing an analytical overview of some previous efforts, the present contribution introduces a novel two-step approach towards bottom-up urban energy computing, involving a reductive phase and a re-diversification process. The reductive phase is performed through an automated process within a GIS platform. The developed process utilizes available large-scale data to generated an energy-relevant representation of the urban building stock. A matrix of energy-influential building characteristics, depicted as aggregate descriptive indicators, is computed based on the generated representation and subjected to Multivariate Cluster Analysis methods for stock classification. The resulting classes are represented through typical buildings, which undergo detailed performance simulation computations. The re-diversification process addresses the loss of diversity due to the reductive method, through employment of stochastic occupancy models and model parametrization. This paper reports on the 38 development of the reductive step, illustrating the encountered challenges and the adopted responses.