Comparison of Sr-90 and Y-90 concentrations in groundwater predicted by four different models

Presented on Thursday, December 6, 2018
Sohan Chouhan, J Miller and L Wojciechowski, Canadian Nuclear Laboratories, Chalk River, ON, Canada

Canadian Nuclear Laboratories recently completed a study comparing modelling software used in performing environmental risk assessment of its Sites. This study assessed contaminant migration through a shallow sandy aquifer on the Chalk River Laboratories site.

Sr-90 and its daughter Y-90 radionuclides were considered in this comparison. For this scenario, it was assumed that a one-year pulse of 32 TBq of Sr-90 was released over an area of 115 m2 at the water table, thus transport in the unsaturated subsurface zone was not considered.

The groundwater transport path was assumed to consist of two parts in series. The first part was 140 m through silty, very-fine sand, and the second was 310 m through fine-medium sand. During transport, the radionuclides were assumed to undergo one-dimensional advection, three-dimensional dispersion, and linear equilibrium sorption. The groundwater was assumed to discharge to a pond and to a well located 1 m upstream from the pond. The well was assumed to be centred on the contaminant plume and did not significantly alter the natural groundwater flow.

The plume ground water concentrations were predicted using RESRAD-OFFSITE Version 3.1, AMBER-LI602, SYVAC3-CC4 Version CC4.09.1, and ECOLEGO-EXERCISE2. The peak of predicted well and pond concentrations of Sr-90 varied from 319 Bq/L predicted by RESRAD-OFFSITE to 1090 Bq/L predicted by SYVAC3-CC4, with ECOLEGO-EXERCISE2 and AMBER-LI602 predictions being in the middle. The peak arrival times varied from 140 years predicted by ECOLEGO-EXERCISE2 to 180 years predicted by RESRAD-OFFSITE, with AMBER-LI602 and SYVAC3-CC4 predictions being in the middle. Similar prediction variations were found for Y-90. Broad reasons for differences in the predictions were identified. This comparison provides an excellent estimate of the magnitude of uncertainties in model predictions for this scenario.

Progress made from work currently being done on including predictions using Visual MODFLOW Flex into the comparison may also be discussed.



Sohan Chouhan
Canadian Nuclear Laboratories, Chalk River, ON, Canada
TBA

J Miller
Canadian Nuclear Laboratories, Chalk River, ON, Canada
L Wojciechowski
Canadian Nuclear Laboratories, Chalk River, ON, Canada
NGWA may only post those brown bag sessions, Webinars, event sessions, and like for which we have obtained copyright permission from the presenter. Furthermore, by accessing or downloading any of these items, you agree they are for your own personal use and may not be disseminated by any means to others via any medium. Click here to read NGWA's proprietary legend and disclaimers before proceeding.