Recently, the interest in urban-scale modelling of energy transfer phenomena has markedly increased, due to multiple contributing factors. The steadily rising prominence of cities as humanity's habitant is one. Another factor is a deeper understanding of cities as highly nested constituents with complex behavioral traits – also with regard to the dynamics of energy generation, distribution, storage, and use. As with other complex problems, computational modelling of the pertinent processes can provide effective support for both systems analysis and decision making. Specifically, the conception and evaluation of energy performance improvement strategies for the urban environment requires reliable predictions of the spatial and temporal distribution of energy demand and supply. This implies the need for versatile modelling environments that can facilitate spatiotemporally detailed energy-related district-level inquiries, pertaining, for example, to candidate energy efficiency measures. In this context, the present contribution elaborates on a novel urban energy computing environment, which intends to accommodate three essential requirements: i) Realistic representation of the dynamics and diversity of urban-level microclimatic boundary conditions requirement; ii) Realistic representation of internal (occupancy-related) processes in buildings; iii) High-resolution numeric simulation of underlying transient heat transfer phenomena. To satisfy these requirements in a computationally feasible manner, a specific hourglass paradigm is deployed. Thereby, a reduced representative sample of buildings in the pertinent urban analysis domain is generated. This reductive step allows to use, instead of simplified and reduced order algorithms, detailed (high-resolution, transient) numeric modelling, while circumventing the problem of massive data requirements and extensive computational loads. The reductive step, however, inadvertently results in some loss of complexity and diversity. The proposed hourglass paradigm thus involves a second step toward targeted recovery of lost diversity via probabilistically based variation of multiple input data sets for extensive parametric simulations. As such, the resulting simulation-assisted urban energy decision support environment facilitates the analysis and comparative evaluation of various energy-related intervention scenarios pertaining to macro and microclimate conditions, demographics and behavioral issues, physical and technical aspects of the buildings, and urban morphology.