Lessons learned from multiple fidelity modeling of ground interferometer testbeds

The MicroPrecision Interferometer Testbed (MPI) at JPL is a dynamically and dimensionally representative hardware model of a future spaceborne optical interferometer.(1) Over the past few years, several models of MPI have been created. These include detailed, high-fidelity models of MPI and several lower-fidelity models. These models were meant to answer two basic questions: (1) "Does current modeling methodology allow accurate models of highly complex opto-mechanical systems such as the MPI testbed?" and (2) "Given a valid modeling methodology, how much model fidelity is needed in models to accurately predict performance?". In order to answer these questions, four models of the MPI testbed were created; each with a unique optical and structural model fidelity. This paper reviews results obtained from these models. It compares disturbance transfer function predictions from three of the models with measured disturbance transfer functions from the hardware testbed. (The fourth model, not discussed here, quantifies the effects of model-updating using both modal and in-situ component testing.) Results suggest that it is possible to build a highly accurate high-fidelity model, thus validating the modeling methodology. With lower fidelity models, meaningful model prediction errors exist when simple models are used to represent the complex opto-mechanical system. However, modest increases in model fidelity can lead to significant improvement.