Building design: Reducing calculation time for simulation and optimization studies
The holistic approach, which considers the building as a whole and over a long period, could represent a solution for performing efficient simulation and optimization studies and is currently raising interest in the scientific community. This type of global approach to building modelling takes into account the strong interactions between the envelope, the systems, the environment and the users. Its main drawback is that it requires computation times that are too long, if not unfeasible, when using detailed models in dynamic regime over long simulated periods. The challenge is to reduce this computation time. This is usually achieved by using simplified models instead of detailed building models, but this can significantly affect the representativeness of the case studies. Furthermore, simplified models may not be applicable for large complex models, as they require a lot of training data to be built.
Work at INES is exploring another path, based on reducing the simulated period, instead of the model itself. The objective is to define a sufficiently short and representative sequence of days to determine performance with the full dynamic model and then extrapolate it to the full year.
This work has led to the development of a day selection methodology called the Typical Short Sequence (TSS) algorithm to generate reduced sequences of typical days that can be applied to detailed models, despite their level of complexity, in dynamic simulations. The algorithm, applied on a detailed building model, leads to much faster simulations while obtaining results very close to the annual results.
An optimization approach, called OptiTypSS, is also evaluated. It uses the reduced sequences generated by TypSS to accelerate heavy multi-objective optimisation studies.
The method allows obtaining results very close to the optimal obtained from simulations over a full year. Its calculation time still needs to be improved.
This work was the subject of Hasan SAYEGH's thesis (https://tel.archives-ouvertes.fr/tel-03219964), co-supervised by the University of Savoie Mont Blanc (LOCIE-CNRS) and the CEA at INES, within the framework of the OREBE project financed by the Auvergne Rhône-Alpes Region.