Container failure by general corrosion is likely to result in sufficient opening of the container surface to allow substantial advective release of radionu­clides. The rate of release of radionuclides by advective release is expected to be higher by several orders of magnitude than diffusive release that may occur through tight cracks or small pits. Therefore, underestimating the general corrosion rates because of the uncertainties may lead to an inac­curate, delayed and low-magnitude radionuclide release from a failed con­tainer. When the uncertainties associated with the general corrosion rates are random in nature, the general corrosion rates are expressed in a uniform distribution, either in a linear scale or in a log scale, depending on charac­teristics of the uncertainties. Figure 7.4 is an example output of the failure probability of a carbon steel container with time, using the range of general

corrosion rates sampled from a log normal distribution extracted from Fig. 7.2 (David et al., 2002; Jung et al., 2011). The radionuclide release at a given time will begin from a finite number of containers that failed, from calcula­tions using the probability of failure and the total number of containers.

Container failure by localized corrosion may also limit the radionuclide release because of restricted flow through the small perforations due to the pits. Pit diameters were from micrometers to millimeters (0.4 microinch to milliinch) and pit density is 0.1-100/cm2 (0.6-645/inch2) as shown in Fig. 7.5 from selected metals and aqueous environments (Szklarsksa-Smialowska, 1986). In addition, the pits are usually filled with corrosion products or solid precipitates from groundwater. Therefore, radionuclide release through the restricted area is likely to be diffusive, which is generally slower than advec — tive release.