Numerous experimental and theoretical contributions in the past have stressed the detrimental effect of fractures in the generation of surface laser damage sites in fused silica illuminated at 351 nm. However, two very important steps lack for the moment on the way towards a scientific understanding of the role of fractures. 1. a physical model must be developed to predict damage events starting from real defect sites 2. a reproducible measurement must be obtained and compared with calculations.Here we present the theoretical work realized to reach the first goal. Contrary to previous discussions on fractures, the electromagnetic configuration is calculated in the case of a real material, with electronic surface states, bulk defects, and defects dynamics. Due to electromagnetic field enhancement in the fracture, surface defects absorb a sufficient part of laser energy, able to heat silica above the vaporization temperature. This is the initial event that triggers production of more excited states during the pulse, and steep increase of temperature and pressure fields. Comparisons with available experimental results are positive. Calculated fluences of damage initiation are very near those of measured events on engineered fractures, or on real defects in polished samples.