BIM-4-SIM -Urban Buildings Generation and Modeling for Thermal and Energy Simulations
Due to its health, energy, and economic consequences, the degradation of the quality of urban environments, especially due to overheating, has become one of the major issues of urban development. To bring some solutions to this problem, physical modeling and simulation of urban areas (including buildings and outdoor environments) show great potential. However, the availability and the construction of urban models based on an analysis of geometric shapes is a big issue limiting the possible simulations and the quality of their results. For instance, airflow simulation generally requires input watertight models and identification of plant canopies, while thermal and energy modelings need to know the nature of construction materials, the properties of urban surfaces, and often intra-building breakdowns, data that are generally unknown.
Due to the lack of availability of urban models that meet these needs, most of the current research in urban simulation uses simplified models built manually with existing databases (IGN BD Topo, Open Street Map, CityGML databases, …) by using specific 3D modelers. This construction process is time-consuming and limits the possibilities of carrying out numerous simulations. It thus restricts the possibilities of simulation studies and the reproducibility of experimentations with different parameters. A second major bottleneck is the lack of data available to reconstruct the interior and exterior of buildings at a large scale. Several solutions exist to reconstruct the external shape of the buildings, using for instance Lidar scans, but data representing both the exterior and interior of buildings are very rare.
Our goal in this project is to answer these limitations through three scientific objectives:
- (1) define a method that allows generating automatically 3D building models based on many different parameters;
- (2) propose several algorithms based on open data, crossing multiple sources, to extract automatically the parameters required by our method of building generation;
- (3) implement dedicated model extraction algorithms that will be used as input of software for physical simulation; propose extractions at different scale to improve the quality of physical simulations in terms of parameterization, reliability and results valorization towards society.
One originality of our approach is to consider the problem of 3D building representation for physical simulation not from the point of view of reconstruction from data but from a generation problem where the building will fit the best possible data available. Such an approach solves the problem of lack of data. The second main originality is the use of a topological data structure, a combinatorial map that describes precisely the topology and the geometry of buildings while allowing for construction and modification operations. The third originality is the decorrelation of the data analysis task and the model generation one that allows to work on the two aspects in parallel and limits the risk of failure of one task in case of problems on the other one.
