Discrete adjoint constrained design optimization approach for unsteady transonic aeroelasticity and buffet
We present the details of two primary research topics in the area of aeroelastic optimization that we are currently working on. The first topic is a new discrete adjoint aeroelastic design optimization solution methodology, which has a computational cost that is independent of the number of structural degrees-of-freedom of the aeroelastic model. In recent years, we have developed a discrete adjoint nonlinear harmonic balance frequency-domain unsteady aeroelastic design optimization methodology, which is able to compute aeroelastic geometric design gradients, that can then be used with an optimization solver to compute an alternate geometric design, which will delay the onset of flutter. This methodology has proven to be robust, however it can be computationally costly since the computational cost is proportional to the number of structural degrees-of-freedom. The new methodology we present does not have that issue, and only a single adjoint solution computation is required when design gradients are required by whatever optimization algorithm one happens to be using. The reason this is possible is that we have recently developed a new nonlinear harmonic balance frequency-domain based aeroelastic solution technique where both the aeroelastic solver and computational flow solver can be marched to convergence simultaneously. With our original aeroelastic solution methodology, we run the aeroelastic solver and flow solver sequentially, one after the other. With the ability to march the aeroelastic and flow solvers simultaneously, we automatically satisfy the constraints that the aeroelastic residual and flow solver residual are zero once the combined aeroelastic solver and flow solver is converged. The second topic discussed in this abstract are new transonic aeroelastic flutter and transonic buffet design optimization results that are computed using a constrained optimization solver. Previously, our transonic aeroelastic flutter and transonic buffet design optimization results were computed using an unconstrained optimization solver. With the new constrained optimization solver, we are again computing transonic aeroelastic and buffet optimized geometric designs, which best delay transonic aeroelastic flutter and transonic buffet onset, respectively, however now with the additional constraint that the steady lift-to-drag ratio of the new design must be greater than or equal to the steady lift-to-drag ratio of the nominal unmodified geometric design of the configuration. We present details and results for both of these topics.
