Formatted Title

Applying a Stochastic Approach to Developing a Water Balance in a Data-Limited Environment

Background/Objectives

Between 1991 and 1996 the U.S. Department of Energy designed and constructed the Rifle, CO disposal cell to encapsulate uranium mill tailings sourced from the nearby former New and Old Rifle uranium mill processing sites. The disposal cell was constructed with a soil and rock riprap cover designed to limit radon emission and precipitation infiltration and is unlined to promote fluid drainage into bedrock. Because the disposal cell was constructed within a drainage feature, a high-density polyethylene (HDPE) liner was installed along the downslope toe to contain any phreatic surface that were to develop until it could drain into the underlying bedrock. Design simulations predicted that fluid levels within the cell would initially rise due to gravity drainage of tailings fluids present within the tailings at the time of entombment, but would peak before exceeding the top of the liner and decline once the rate of gravity drainage from the tailings dropped below the infiltration capacity of bedrock.  As a precaution, a leachate collection system consisting of three vertical standpipes was installed to facilitate leachate level monitoring and dewatering in case disposal cell fluid levels were to approach the top elevation of the liner.  Disposal cell fluid levels have continued to increase and pumping from the leachate collection system has been unsuccessful at reversing water level rise. Water levels are now threatening to exceed the top of the HDPE liner. Evaluation potential was limited by data availability; the only data available was temporal water levels from standpipes during periods where pumps were not operating, which, based on cell bathometry, were converted to fluid volumes, and standpipe temporal extraction rates.  Design documents and accompanying reports provided ranges of expected infiltration rates through the cover and into bedrock, and uncertain estimates of initial tailings moisture content.  To overcome data limitations, a stochastic water balance was used to characterize the combinations of inflows and outflows that reasonably matched the observed temporal increases in storage and those combinations were used to predict extraction rates and reductions in cover infiltration rates required to reduced disposal cell fluid levels.  

Approach/Activities

The water balance included recharge through the cover system, release of fluid from initial moisture in the tailings, drainage into the underlying bedrock, and removal of water via the leachate collection system. Reasonable ranges of inflow and outflow rates were established and randomly sampled to create one million unique realizations. Temporal storage estimates for each realization were evaluated against observed temporal storage trends in the disposal cell, and realizations that failed to replicate the observed trends were rejected. 235 realizations were retained and used to evaluate the probability that various dewatering and reductions in cover infiltration rates would permanently reduce disposal cell fluid levels below the top of the liner.

Results/Lessons Learned

The retained realizations provided plausible bounds on the ranges of water balance inflows and outflows. These final ranges suggested that rising water levels in the disposal cell could be attributed to recharge through the cover consistently exceeding discharge into bedrock. Forecast simulations determined that increasing the annual extraction rate from an average of 0.3 gallons per minute (gpm) to an average of 4 gpm for a 2-year period would result in water levels declining below the top of the HDPE line in 94% of realizations. Another forecast simulation demonstrated that after a period of increased extraction, a reduction of infiltration through an evapotranspiration cover would eliminate the need for perpetual pumping and maintain water levels below the top of the liner. The water balance provided crucial data that facilitated creation of a numerical flow model used for design and evaluation of dewatering designs and cover modifications.