No label defined (Q4938237)

The paper deals with a crack shape sensitivity analysis based on Galerkin boundary integral equation (BIE), domain differentiation and adjoint variable. The shape sensitivities can be evaluated using either a direct differentiation or the adjoint variable approach. The reviewed paper has being focused on the latter. The adjoint variable approach needs to solve only one new boundary value problem per integral functional, whatever the number of shape parameters. In connection with BIE formulations alone, the adjoint variable approach has been successfully applied to many shape sensitivity problems. It relies critically upon the possibility of formulating the final analytical expression for the shape sensitivity of a given integral functional as a boundary integral which involves the boundary traces of the primary and adjoint states. However, this step raises difficulties when cracks (elasticity) or screens (acoustics) are present: non-integrable quantities associated with crack front singularity of field variables arise in the process. The reviewed paper purports to show that this difficulty can be avoided by formulating, in the Lagrangian, the direct problem constraint as a symmetric Galerkin BIE. As a result, explicit boundary-only expression for sensitivities are obtained for shape perturbations of either voids or cracks, in linear acoustics and elastodynamics. For the entire collection see [Zbl 0924.00038].