ACCELERATING WASTE SITE CLEANUP AT THE SAVANNAH RIVER SITE USING GEOSTATISTICS

In recent years, the push to accelerate cleanup at Department of Energy (DOE) sites has become the latest trend in environmental restoration. In order to accomplish this, site characterization must be quick and efficient. There are many tools available to expedite characterization, but one of the most effective tools is the use of statistics in sample planning. At the Chemicals, Metals, and Pesticides (CMP) Pits waste site at the Savannah River Site statistical methods were used to expedite site characterization while maintaining data quality and cost effectiveness.

Determining how many samples are required to complete characterization of a waste site is typically subjective and can involve large numbers of samples, long durations of time, and large budgets. By using DOE’s Visual Sample Plan (VSP) statistical software the efficiency and effectiveness of the CMP Pits characterization were improved. VSP allows users to choose from a variety of statistical methods to best determine the number of samples needed to properly characterize a site. A site map can be loaded into the VSP program and statistical parameters can be tailored to the particular site. Once the appropriate number of samples has been determined, VSP will randomly generate locations, which can be exported to a GIS program and incorporated into the sample and analysis plan.

At the CMP Pits, VSP was used to plan verification sampling for an area where contaminated soil had been removed. By performing statistically based sampling of the area rather than traditional grid sampling, the number of samples was reduced by 50%, thus reducing the cost of sampling by $8,000. Additionally, when the sampling plan was presented to federal and state regulators, the statistical basis for the sampling plan was accepted and required no additional sampling for verification of the remedial goal objectives.

Using statistics for characterization planning incurs no additional labor cost, is scaleable for project needs, reduces characterization costs and time, results in statistically valid sample plans with reduced subjectivity, and produces increased regulator confidence.