Laminar separation from the suction side of low-pressure turbine (LPT) blades can significantly degrade engine efficiency. If laminar separation could be controlled, aerodynamic performance could be maintained over a wider operating range. Also, new and more aggressive stage designs with reduced solidity (less blades) would become possible. One example is the new L1M LPT blade, which was designed to allow for an increased aerodynamic loading by applying active now control (AFC) in a continuous fashion. Motivated by this and other success stories where open-loop flow control was shown to result in dramatic performance improvements, the general focus has shifted towards closed-loop control which promises even greater gains and a greater robustness and capability to adjust to changing operating conditions. However, a fully satisfactory methodology for designing robust and efficient closed-loop controllers for fluids problems has not been devised yet. This paper summarizes different approaches for investigating control of flow separation from the L1M blade using 2-D numerical simulations. A parameter study with open-loop control by harmonic wall normal blowing upstream of the separation was conducted to determine the optimum forcing parameters. We believe that by exploiting flow instability mechanisms the flow control can be made more efficient. It was also demonstrated how a proportional differential (PD) controller with self-adjusting parameters can be employed successfully for closed-loop control. Even better closed-loop controllers will likely become possible if a real-time prediction of the now dynamics over a reasonably broad parameter range became available. Galerkin models perform well for one given operating point but do not generalize well. Here, neural networks were explored for making real-time predictions of the unsteady separated L1M now field. The resulting models are shown to be both accurate and robust and to generalize well.