Visualization is an omnipresent technique to present data and concepts. Conducting user studies is a state of the art to evaluate the readability and ergonomics of visualizations. However, the preparation, conduction and analysis of user experiments is very time consuming. Modeling of visual search strategies is not yet a widespread tool for visualization design. This paper presents results of ongoing work in developing a simulation framework providing techniques to set up and run cognitive models for visual search prediction. Model designers can set up declarative knowledge about visualizations and the framework provides algorithms for a cognitive simulation. We present an implementation of this framework in C# and ACT-R, and demonstrate how we set up an ACT-R model for the simulation of a bar chart reading task. Finally, we discuss strengths and limitations of the framework and present future work.