Sensitivity analyses provide information how process responses vary with model parameters including material, shape, contact, boundary and initial conditions. This information is directly applicable in the design procedures as well as a part of gradient-based algorithms for inverse analyses and optimisation.
C3M has invested considerable resources in the development of algorithms for analytical evaluation of sensitivities such as direct differentiation and adjoint method which are much more efficient and robust than finite difference approximations. Furthermore, C3M is developing advanced approaches to evaluate sensitivities even when numerical responses are non-smooth and contain relatively high degree of noise. |
|
