We discuss the relevance of object-orientation in the field dynamic system simulation. Difference and differential equations are the well known methods to describe structure and behavior of such systems. Traditionally, these descriptions are directly translated into computer programs. This approach, though computationally efficient, shows several drawbacks: The semantics of a model buried, model and program reuse is limited. We show how the application of the object-oriented paradigm can help overcome these drawbacks. The work is based on an environment which supports the classical concept of coupled models as well as a novel "building block", the so-called array model consisting of a series of structurally compatible components (i.e., instances of the same model class subclasses thereof) in parallel corresponding to the individuals of a set to be modeled. As the number of components in an array model may vary over time, this concept allows for dynamic adaptation of a model?s structure.