“…The calculation of softening models such as those defined in ''The case of softening damage'' section over a domain goes through a critical point where a loss of ellipticity, i.e., a loss of uniqueness of the solution (Needleman, 1988;Peerlings et al, 1998;Pijaudier-Cabot and Benallal, 1993) occurs and leads to pathological mesh dependencies (Desmorat et al, 2010;Ladeve`ze et al, 2000). Several remedies exist for damage models, including nonlocal formulations (Desmorat et al, 2010; The delay-effect method First, in order to have a damage variable with an order of magnitude of 1, let us rewrite eqs.…”