Automating building energy system modeling and analysis: An approach based on SysML and model transformations

A methodology to automate building energy model generation and system analysis is proposed. The contribution of this methodology is to extend the formal model transformation framework toward building energy system modeling, thus enabling efficient and flexible system synthesis through model decomposition and recomposition. It builds on SysML metamodeling and Triple Grammar Graph based model transformation that takes in the descriptive architecture model and outputs the corresponding TRNSYS model in a standardized fashion, thus greatly reducing the potential of heuristic modeling uncertainty. The automation is also extended to the analysis of the synthesized system; thereby human intervention becomes minimized, which expedites decision-makings at various steps of the model-based design exploration. The automation is applied to a special design problem of Model-based Predictive Controls that may contain the heuristic modeling uncertainty in case of custom modeling, and that also requires very frequent system analyses during development. Illustrative examples demonstrate modeling reliability, as well as enhanced modeling and analysis productivity, compared to the manual process. The enhanced productivity even allows extra explorations. If a designer were to create a TRNSYS model by hand, it would most likely be economically feasible to carry out only a few types and rounds of analyses for only a few system configurations at best. With the automated TRNSYS model generation and system analysis, designers can efficiently perform a much larger number of detailed analyses, which would result in an improved exploration of both the design space and the selected alternatives.