A generic framework for the simulation of transient dynamics in nonlinear aeroelasticity is presented that is suitable for flexible aircraft maneuver optimization. Aircraft are modeled using a flexible multibody dynamics approach built on geometrically nonlinear composite beam elements, and the unsteady aerodynamics on their lifting surfaces is modeled using vortex lattices with free or prescribed wakes. The open loop response to commanded inputs and external constraints is then fed into a Bayesian optimization framework, which adaptively samples the configuration space to identify optimal maneuvers. As a representative example, the proposed approach is demonstrated on a catapult-assisted takeoff. The specific modeling challenges associated to that problem are first discussed, including the effect of aircraft flexibility. An optimality measure based on ground clearance and wing root loads is then defined. It is finally shown that the link that ramp-length constraints introduce between acceleration, release speed, and wing root loads is the main driver in the optimal solution.