DD-DIC : a parallel finite element based digital image correlation solver
Résumé
The computational burden associated to finite element based digital image correlation methods (FE-DIC) is mostly due to the inversion of global FE systems and to global image interpolations. On the contrary, subset based approaches require only subimages, and allow solving small independant systems in parallel. A variable separation technique was recently proposed that alleviate mesh constraints [Passieux and Périé IJNME 12] in FE-DIC. However, in digital volume correlation, the question of the interpolation of the images becomes a real issue [Perini et al. Strain 14]. For that reason, a non-overlapping dual domain decomposition method is proposed to rationalize the computational cost of high resolution FE-DIC measurements when dealing with large datasets [Passieux et al. IJNME 15]. It consists in splitting the global mesh into submeshes and the images into subimages. The displacement continuity at the interfaces between subdomains is obtained iteratively by using a preconditioned conjugate gradient. It will be shown that the method combines the metrological performances of finite element based DIC and the parallelisation ability of subset based DIC methods.