Our new publication focuses on the derivation of an adjoint structural stress and vibration solver using algorithmic differentiation. Both derivations utilize a combination of symbolic and algorithmic differentiation techniques and enable gradient calculations at a cost independent on the number of inputs.

This post introduces the basic workflow of a CAD-based adjoint multidisciplinary optimization framework, which has been applied to perform a CAD shape optimization of a radial turbine to improve aerodynamic efficiency, while keeping maximum von Mises stress values below a defined threshold.