Level of detail in dynamic simulation models – a comparison for multi-family houses

To achieve the set climate goals, renovations andoperational optimizations in the residential sec-tor are necessary. In this context, building per-formance simulations have proven to help esti-mate the CO2 saving potential of different renova-tion measures. Various modeling approaches ex-ist that differ in their level of detail ranging fromcomprehensive physical descriptions to simplifiedrepresentations. While literature covers compar-isons of dynamic simulation models for single-family houses and non-residential buildings, a com-prehensive analysis for multi-family houses is miss-ing. Besides a different architecture, one distinctcharacteristic is their high diversity in occupancypatterns. Especially due to diverse user behaviorand different resulting temperature set points, heatexchange between floors and flats in multi-familyhouses are of particular interest. This study com-pares models with different levels of detail in open-source tools (EnergyPlus and Modelica) for predict-ing heat demand in multi-family houses, taking intoaccount inter-zonal heat exchange. Detailed build-ing models in EnergyPlus and Modelica performsimilarly in terms of accuracy, but EnergyPlus out-performs Modelica regarding the simulation time.For models with different levels of detail in Model-ica, the results highlight trade-offs between accu-racy and computational efficiency. Reduced-ordermodels outperform high-order models in terms ofsimulation time (up to 98 %-99 %) but at the ex-pense of accuracy in estimating the annual heatdemand on flat level (13 %-46 %)