Visual Analytics is one effective way to improve the understanding of large and complex datasets through the systematic combination of interactive visualizations and automated analysis techniques. Time-oriented data are highly relevant to many application fields of Visual Analytics but time and time-oriented data do also have a complex semantic structure involving design aspects such as granularities of time, different time primitives, and indeterminacy. Still, most existing software frameworks for visualization and Visual Analytics support only a subset of these design aspects to satisfy concrete application demands. However, for prototyping in basic research on Visual Analytics methods, a software framework is needed that supports the design aspects of time-oriented data in a systematic, theory-driven way. Tackling such need, this work investigates how a software framework can support Visual Analytics of time-oriented data in an expressive and efficient way. Its outcomes comprise substantial parts of the conceptual software architecture and the prototypical implementation of TimeBench. In particular, this work focuses on TimeBenchs data structures and import/export functions. The software architecture and implementation were designed based on established software design patterns for visualization such as the Proxy Tuple pattern. The resulting data structures are conceptually based on two interlinked direct acyclic graphs and implemented on top of the relational data tables provided by the prefuse framework. For evaluation, the work presents the reimplementation of two complex visualization techniques and two development case studies. These results demonstrate that TimeBench is useful for Visual Analytics prototyping with a focus on time-oriented data well.