The 25.4 m Giant Magellan Telescope (GMT) consists of seven 8.4 m primary mirror (M1) segments with matching segmentation of the Gregorian secondary mirror (M2). The GMT will operate in four basic optical correction modes, Natural Seeing (NS), Ground Layer Adaptive Optics (GLAO), Natural Guide Star Adaptive Optics (NGAO) and Laser Tomography Adaptive Optics (LTAO). Each of these modes must deliver a specified combination of image quality, field of view, and sky coverage over a range of environmental conditions.
With a double segmented mirror configuration, even in the simplest of the correction modes the GMT includes over one thousand controllable degrees of freedom. Exogenous and internal sources of disturbances and noise over these degrees of freedom will limit the image quality. The different ranges of motion and bandwidth of the different degrees of freedom enable a cascade correction of the wavefront error, successively rejecting global to local disturbances. This frequency and spatial separation allows allocating the disturbances in stages, considering the residuals of the low spatial and temporal corrections as the disturbance for the high order corrections.
While a first approach can consider the analysis of systems in isolation in order to allocate coarse budgets, a complex control system such as that of the GMT requires a Dynamic Optics Simulation (DOS) to account for the real interactions between the controlled plant and the controllers. For example, some control loops such as the M1 figure control system will have an update rate of only 0.03 Hz, while the Adaptive Secondary Mirror (ASM) will be updated at 1kHz . The DOS is an end-to-end simulation environment that brings together optics, finite element models (FEM), mechanical motions, surface deformations and control models applied to the GMT main optics. At the center of the DOS there is an optics propagation module with both geometric ray tracing and Fourier propagation capability. The dynamic response of the telescope mount and the large M1 segments has been modeled by applying Craig-Bampton reduction analysis to finite element models. These reductions have been reordered in a second order form, allowing higher computational efficiency than traditional state space models. Each M1 segment is controlled by an array of 330 actuators with realistic precision, noise and discretization errors. The structural dynamics model can be used in time domain simulations that account for all the non-linear effects of actuators and sensors, or in a linear frequency domain model to run more efficiently stochastic analyses.
A high resolution Computational Fluid Dynamics (CFD) model has been developed for simulating unsteady turbulent flow over the optical system. These simulations provide unsteady pressure fluctuations over the main optics and effects of varying index of refraction in the optical path for different operating conditions. These quantities are subsequently used for estimating wind induced image jitter and thermal (dome and mirror) seeing by applying the combined structural, control, and optical models described above.
The DOS allows GMT to understand the sensitivity of image quality to any of the thousands of parameters of our plant and control system., Due to the cascade layers of control loops, DOS allows specifying design parameters without over-constraining the solution space.