Systeem-gebaseerd simulatie van energie stromen

Decisions in early design phases highly influence the overall consumption of energy and resources throughout the lifecycle of buildings. However, there are no computer-aided methods adequate for the Level-of-Development (LoD) in these phases which allow a simulation and analysis of the overall resource demand of the planned building. Available monolithic simulation platforms for energy processes in buildings are suited only for later phases because of the high information demand and modelling effort. These platforms do not allow for exploiting the potential of early design phases.
The project, which provides a chief analysis method for the researcher group for the evaluation of buildings in early design phases, aims at the development of a modelling and simulation approach that allows for a simple, quick and sufficiently exact estimation of the operational and embedded energy demand for a multi-LOD model. This approach needs to manage incomplete and fuzzy information and, if necessary, to make further assumptions or to request further information. Additionally, it should deliver an estimation of the reliability of the results that includes information on their accuracy and which information could improve them.
Methodical basis of the project is the paradigm of parametric object-oriented and flow-based systems modelling and simulation of buildings. The system model to be developed shall have two chief modelling levels: First, the system flow modelling and, second, a modelling of parametric interdependencies. The systems flow modelling captures the relevant energy and material flows, if necessary, in a dynamic way. The model of parametric interdependencies stores the calculation methods of the flow quantities. For real-time design assistance, surrogate models are developed on the basis of the second-level model that captures parametric interdependencies. Mathematical-statistical surrogate models (black box) as well as simplified physical models (white box) and combinations of both methods (grey box) are tested in order to develop this model. The necessary representation of fuzziness takes place on the basis of stochastic approaches. For the tracking of effects of changes as well as fuzziness, methods of error propagation become part of the systems model. For the synchronisation with the multi-LOD product model, the project includes the development of rule-based and knowledge-based transformation methods of the systems model. Basis for these rules are graph transformations that serve the purpose of the initial derivation of the systems model from the product model and of updating. In this context, an examination of the possibilities of an adaptive partial recalculation takes place to allow a reaction to selective changes and detailing processes as well as an effective analysis of variants.