System integration in design and operation as a means to progress towards cost-effective Net Zero Energy Buildings requires different modeling approaches than the ones offered by today's building energy simulation programs. Based on his research and on advances in other industrial sectors, the presenter will show how hierarchical object-oriented computer models that describe physical identities instead of variable assignments facilitate multi-physics modeling, modeling of large dynamical systems and provide a path towards embedded computing within the design process and building operation. He will discuss why next-generation building system simulators should be built using hierarchical objects in which the model formulation is separated from the numerical solution processes. He will show how this will facilitate the integration of multi-physics models to analyze phenomena which are characteristic for high performance buildings, how it will support concurrent engineering design processes and how it will facilitate the reuse of models throughout the building life cycle. He will introduce the Modelica modeling language, which is considered by a large system simulation community to become the next-generation multi-physics modeling language. He will discuss his experience in developing Modelica models for building energy system simulation and optimal control. He will highlight opportunities and research needs and outline a strategy for developing a next-generation building simulation program.