[en] Simulations were conducted to analyze the sensitivity of the engineered barrier system (EBS) release rate to alternative conceptual models of the advective release from waste packages under dripping fractures. The first conceptual model assumed that dripping water directly contacts the waste form inside the 'failed' waste package, and radionuclides are released from the EBS by advection. The second conceptual model assumed that dripping water is diverted around the 'failed' waste package (because of the presence of corrosion products plugging the perforations) and dripping water is prevented from directly contacting the waste form. In the second model, radionuclides were assumed to transport through the perforations by diffusion, and, once outside the waste package, to be released from the EBS by advection. The second model was to incorporate more realism into the EBS release calculations. For the case with the second EBS release model, most radionuclides had significantly lower peak EBS release rates (from at least one to several orders of magnitude) than with the first EBS release model. The impacts of the alternative EBS release models were greater for the radionuclides with a low solubility (or solubility-limited radionuclides) than for the radionuclides with a high solubility (or waste form dissolution-limited radionuclides). The analyses indicated that the EBS release model representing advection through a 'failed' waste package (the first EBS release model) may be too conservative in predicting the EBS performance. One major implication from this sensitivity study was that a 'failed' waste package container with multiple perforations may still be able to perform effectively as an important barrier to radionuclide release. (author)