Measured swelling, hydraulic and thermal properties of MX-80 bentonite: Distinguishing between material variability and measurement limitations

The Canadian Nuclear Waste Management Organization (NWMO) is investigating Sedimentary and Crystalline media as potential hosts for a Deep Geological Repository (DGR) for permanent isolation of used nuclear fuel and fuel wastes. In NWMO's concept, densely compacted bentonite clay is used to surround the used fuel containers and to backfill the placement rooms. Development of sealing concepts and performance prediction models requires many materials properties inputs. These inputs are important to the evaluation of hydraulic conductivity ( k ), swelling pressure ( P s ) and heat transfer. Evaluation of these and other properties is complicated by factors such as unidentified variability in material composition (mineralogy) and differences in test methodologies. In order to eliminate the effects of technique, data obtained using similar methods are used in this study. This should result in data scatter being more related to materials variability and limitations than measurement method. In an effort to assess the variability in the reported Ps, k and thermal properties and determine representative values or relationships for input into performance models, sufficiently large databases are needed to allow for statistical evaluation for a full range of potential repository conditions. Unfortunately, in many cases these databases are not extensive, particularly for materials in a highly saline environment. This paper presents statistical evaluations of the relationships between density, k , P s and thermal properties using a database generated from literature sources for a well-known bentonite (MX-80) under freshwater and saline porefluid conditions. Also included is previously unpublished work by NWMO where repository-relevant groundwaters were used in testing. To limit uncertainty, the data used were obtained from sources reporting use of MX-80 bentonite and providing at least basic mineralogical information. From these data, best-fit, confidence and prediction bands and intervals were generated.