Comparison of Results for Quarter 1 Surface Water Split Samples Collected at the Nuclear Fuel Services Site, Erwin, Tennessee

Abstract

Oak Ridge Associated Universities (ORAU), under the Oak Ridge Institute for Science and Education (ORISE) contract, collected split surface water samples with Nuclear Fuel Services (NFS) representatives on August 22, 2012. Representatives from the U.S. Nuclear Regulatory Commission and Tennessee Department of Environment and Conservation were also in attendance. Samples were collected at four surface water stations, as required in the approved Request for Technical Assistance number 11-018. These stations included Nolichucky River upstream (NRU), Nolichucky River downstream (NRD), Martin Creek upstream (MCU), and Martin Creek downstream (MCD). Both ORAU and NFS performed gross alpha and gross beta analyses. The comparison of results using the duplicate error ratio (DER), also known as the normalized absolute difference. A DER ≤ 3 indicates that, at a 99% confidence interval, split sample results do not differ significantly when compared to their respective one standard deviation (sigma) uncertainty. The NFS split sample report does not specify the confidence level of reported uncertainties. Therefore, standard two sigma reporting is assumed and uncertainty values were divided by 1.96. A comparison of split sample results, using the DER equation, indicates one set with a DER greater than 3. A DER of 3.1 is calculated for gross alpha results from ORAU sample 5198W0003 and NFS sample MCU-310212003. The ORAU result is 0.98 ± 0.30 pCi/L (value ± 2 sigma) compared to the NFS result of -0.08 ± 0.60 pCi/L. Relatively high DER values are not unexpected for low (e.g., background) analyte concentrations analyzed by separate laboratories, as is the case here. It is noted, however, NFS uncertainties are at least twice the ORAU uncertainties, which contributes to the elevated DER value. Differences in ORAU and NFS minimum detectable activities are even more pronounced. comparison of ORAU and NFS split samples produces reasonably consistent results for low (e.g., background) concentrations.