PNNL

Abstract

From the mid-1940s through the 1980s, large volumes of wastewater were discharged at the Hanford Site in southeastern Washington State, causing a large-scale rise (in excess of 20 m) in the water table. When wastewater discharges ceased in 1988, groundwater mounds began to dissipate. This caused a large number of wells to go dry and has made it difficult to monitor contaminant plume migration. To identify the wells that could potentially go dry, a first order uncertainty analysis was performed using a three-dimensional, finite element code (CFEST) coupled with UCODE, a nonlinear parameter estimation code. The analysis was conducted in four steps. First, key parameter values were identified by calibrating to historical hydraulic head data. Second, the model was tested for linearity, a strict requirement for representing output uncertainty. Third, results from the calibration period were used to verify model predictions by comparing monitoring wells’ wet/dry status with field data. In the final step, predictions on the number and locations of dry wells were made through the year 2048. A non-physically based model that extrapolated trends at each individual well was also tested as a predictor of a well’s wet/dry status. Results demonstrated that when uncertainty in both parameter estimates and measurement error was considered, the CFEST-based model successfully predicted the majority of dry wells, outperforming the trend model. Predictions made through the year 2048 identified approximately 50% of the wells in the monitoring well network are likely to go dry, which can aid in decisions for their replacement.