In this paper, simulations of 500 open-source CityGML building models for heating demand calculation are presented. Using an archetype-based approach and statistical enrichment, Modelica simulations are carried out using the TEASER tool and Dymola. The authors compare the simulation results against measurement data and investigate the impact of years of construction, building usages and refurbishment states in the context of urban building energy simulations. The individual buildings are assigned with information over the construction years and building functions using the measurement data. The years of construction and building usages are primarily obtained in the context of two other research projects and are used for the archetype assignments of the individual building models. Hence, these are well-informed assumptions by the authors, resp. the industry partner experts. This paper includes an in-depth investigation of individual building archetypes with respect to the years of construction. Firstly, the buildings are simulated using externally acquired construction years and pre-assigned CityGML building functions. The simulation results are compared against the measured primary heating demands. Due to high discrepancies between simulations and measured data, the authors secondly, simulate the building models with different refurbishment states. In this paper, the authors conclude that open datasets are indeed important, however, for energy performance simulations, attributes such as construction years, building usages and refurbishment states are necessary and should be included in the open datasets to present an added value.