Using observations in deposition tunnels to avoid intersections with critical fractures in deposition holes

We evaluate the efficiency of utilising a Full Perimeter Intersection (hereafter denoted FPI) criterion, for identifying discriminating fractures in deposition tunnels and deposition holes. We also evaluate the consequences of using FPI, expressed in terms of the degree of utilisation. This is achieved by means of stochastic fracture simulation consisting of two steps. The first consists of computing intersection statistics between the tunnel and the fracture array. The second step consists of computing the degree of utilisation based on the statistics derived from the first step. We base our simulations upon the DFN models produced for the Laxemar and Forsmark study sites, versions 1.2. We anticipate that these models will mature further within the framework of SKB's on-going site investigations and Site Modelling (versions 2.2 and onwards) and though we here discuss the impact of various DFN parameters on our results, it is beyond the scope of the work presented here to evaluate the validity of the published models. The DFN models have been used with no modifications and accepted as they were published. The uncertainties presented here thus mirror the uncertainties of the DFN models.The consequence of using the FPC has been quantified in terms of degree of utilisation which was judged reasonable. The degree of utilisation is higher in Forsmark as compared to Laxemar due to, mainly; the latter's higher intensity of the steep fracture sets in the studied size interval. The FPC was found insufficient to detect all potentially discriminating fractures. It needed to be complemented and we defined a new criterion, EFPC, to also address large fractures in the immediate vicinity of the tunnel, which remain undetected by tunnel mapping. We here proposed a criterion consisting in discriminating also all fractures that intersect 5 or more canister positions. The use of EFPC decreased the degree of utilisation further, though we still judge it to be reasonable. A substantial reduction of unsuitable canister positions is obtained with the application of the suggested criterion. It is also noted that a substantial number of fractures are erroneously marked as being discriminating, thus implying a 'cost' in terms of degree-of-utilisation. Using the Forsmark and Laxemar repository layouts as example, we showed that site specific application of the criterion combined with respect distances significantly increased the efficiency of the criterion to a presumably acceptable level; only canister positions in Laxemar and 4 positions in Forsmark out of 6,000 were erroneously marked as 'approved' whereas around 182 and 98 deposition holes, for Laxemar and Forsmark respectively, are erroneously accepted if the criterion is not applied. The 'cost' in this case is an increase in the required total deposition tunnel length of 4.7 km and .2 km for Laxemar and Forsmark respectively (using 6 m spacing). As the fractures are cautiously assumed to be completely anonymous features possessing no indications of their size, we anticipate that the majority of the erroneously approved positions will in fact be detected if the criterion is linked to an adequate, qualitative judgement based on site understanding and underground information. Despite the uncertainties in the DFN models, we found the results of the simulations sufficiently encouraging to recommend the Full Perimeter Criterion and it's expansion, EFPC, as a method to identify potentially discriminating fractures