Design Optimization
Reduced Order Modeling
Projection-based reduced order modeling (PROM) can be used in the design optimization process. By approximating a large, computationally expensive simulation with far fewer resources, the design process can be sped up significantly. PROMs can also be used to free up computing resources for other projects, or be used in real-time for controls applications.
Discrete Adjoint Sensitivity Analysis
In applied aerodynamics, there are usually few design responses but a large number of design variables that characterize the aerodynamic shape. In such cases, it is preferable to compute gradients using the adjoint method due to its independence from the number of input design variables and simulation step sizes. The adjoint method can provide sensitivities (i.e. the derivatives of objective or constraint functions with respect to arbitrary number of design variables) of a desired output function of interest. An in-depth understanding of sensitivities can potentially aid in the exploration of design spaces defined by hundreds or thousands of variables, which can lead to the discovery of unique, high-value, unintuitive solutions that would otherwise be difficult, or perhaps impossible, to obtain using traditional design methods.