Comment on ‘‘Seepage into drifts and tunnels in unsaturated fractured rock’’ by Dani Or, Markus Tuller, and Randall Fedors Stefan Finsterle 1 Received 1 December 2005; revised 24 March 2006; accepted 8 May 2006; published 11 July 2006. Citation: Finsterle, S. (2006), Comment on ‘‘Seepage into drifts and tunnels in unsaturated fractured rock’’ by Dani Or, Markus Tuller, and Randall Fedors, Water Resour. Res., 42, W07603, doi:10.1029/2005WR004777. [1] Or et al. [2005] propose an analytical model to estimate seepage into subterranean cavities in partially saturated fractured rock. The model is based on the assumptions that (1) all the water flowing through the matrix is diverted around the opening and thus does not contribute to seepage and (2) all the water flowing through the fractures drips into the opening without any restraint or diversion possibility. Given these two key assumptions, the seepage model is reduced to determin- ing the fraction of the total percolation flux that flows through the fractures and the matrix. This task is accomplished by constructing a composite effective permeability function for the fractured rock, following Or and Tuller [2003]. Neglect- ing the presence of the cavity and assuming vertical, unit gradient, steady state flow and the validity of the Bucking- ham-Darcy law, seepage percentage is determined as the ratio between fracture effective permeability and total (fracture and matrix) percolation flux. The approach is used to estimate seepage into a waste emplacement drift excavated from the fractured tuff formation at Yucca Mountain, Nevada, the proposed repository site for the disposal of high-level radio- active wastes. A comparison is made to synthetic seepage calculations presented by Finsterle [2000], and ‘‘considerable differences’’ were observed in the estimated seepage thresh- old values and seepage behavior. [2] The stated objectives of Or et al. [2005] are ‘‘to present a simple and physically based model for establish- ing estimates of seepage rates into large cavities in fractured rocks’’ and ‘‘to improve the physical processes of flow in unsaturated fractures.’’ In their previous publications, the authors have indeed provided very valuable insights into small-scale physical processes affecting unsaturated flow through fractured rock. However, the model presented here neglects key physical processes and hydrogeological fea- tures affecting seepage into a cavity. In the following, I will comment on (1) the assumptions underlying the proposed seepage model and (2) the authors’ presentation of the seepage model currently used by the Yucca Mountain project, specifically the comparison with results of Finsterle [2000]. It will be demonstrated that the conceptual model presented by Or et al. [2005] can be viewed as inherently inconsistent and oversimplified. The model may thus not be applicable to the drift seepage problem of interest. [3] 1. The key assumption of the proposed seepage model is that the seepage flux from the matrix is zero, and, more importantly, that there is no flow diversion around the opening through the fracture system. In essence, their model considers one-dimensional fracture flow embedded within a two-dimensional nonseeping matrix. The resulting seepage curves are direct consequences of this conceptual model, i.e., the conclusions that fracture flow is fully responsible for seepage and that matrix flow does not contribute to seepage are not results of model calculations but prescribed assumptions. Unfortunately, the justification given for these assumptions is (1) flawed and (2) inconsistent with the subsequent development of the conceptual seepage model. [4] First, the justification is based on a comparison of the ‘‘critical percolation threshold’’ for fractures and the matrix, using the model of Philip et al. [1989]. Without going into details of the calculation, it suffices to point out that the ratio of the seepage threshold for matrix and fracture flow (reported as q* 0 matrix /q* 0 fracture =3 Â 10 À8 ) means that seepage from the matrix is initiated for a percolation flux that is about 7 orders of magnitude lower (not higher, as stated in the first sentence of paragraph 7 of Or et al. [2005]) than the flux needed to induce fracture seepage. This is expected, as the matrix with its low-permeability reaches full saturation at the apex of the drift (Philip’s criterion for seepage initiation) at a much smaller percola- tion flux compared to the fractures. Using the information from Table 1 and Figure 2 of Or et al. [2005] along with the assumed equivalence between the van Genuchten and Gardner a parameter (see Or et al.’s section 3), Philip’s model predicts that some seepage occurs from the matrix and none from the fractures for matric potentials in the range between about À10 4 and À10 kPa, which corresponds to a percolation flux range wider than that shown in Figure 3. On the basis of the authors’ own calculations, this contradicts the assumption they derive from the anal- ysis. While the actual seepage flux from the fractures may be significantly higher than that from the matrix (at least for matric potentials higher than about À10 kPa), the assump- tions of the proposed model are not justifiable based on Philip’s model. [5] Second, in Philip’s model, flow diversion around the opening and the existence of a seepage threshold are direct results of a capillary barrier effect, which is evaluated using a continuum approach as a function of the effective capil- lary drive and the geometry of the opening. For flow in fractures, the authors later reject or ignore the conceptual foundations of Philip’s model, namely, (1) that flow diver- 1 Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, USA. This paper is not subject to U.S. copyright. Published in 2006 by the American Geophysical Union. W07603 WATER RESOURCES RESEARCH, VOL. 42, W07603, doi:10.1029/2005WR004777, 2006 1 of 3