Format

Send to

Choose Destination
J Environ Radioact. 2018 Dec;192:587-591. doi: 10.1016/j.jenvrad.2018.04.019. Epub 2018 Apr 30.

Evaluation of Shiryaev-Roberts procedure for on-line environmental radiation monitoring.

Author information

1
Center for Nuclear Environmental Engineering and Science and Radioactive Waste Management (NEESRWM), Clemson University, Clemson, SC, USA. Electronic address: maraw@clemson.edu.
2
Center for Nuclear Environmental Engineering and Science and Radioactive Waste Management (NEESRWM), Clemson University, Clemson, SC, USA. Electronic address: ayman@clemson.edu.
3
Center for Nuclear Environmental Engineering and Science and Radioactive Waste Management (NEESRWM), Clemson University, Clemson, SC, USA. Electronic address: vblizny@clemson.edu.
4
Center for Nuclear Environmental Engineering and Science and Radioactive Waste Management (NEESRWM), Clemson University, Clemson, SC, USA. Electronic address: devol@clemson.edu.

Abstract

Water can become contaminated as a result of a leak from a nuclear facility, such as a waste facility, or from clandestine nuclear activity. Low-level on-line radiation monitoring is needed to detect these events in real time. A Bayesian control chart method, Shiryaev-Roberts (SR) procedure, was compared with classical methods, 3-σ and cumulative sum (CUSUM), for quantifying an accumulating signal from an extractive scintillating resin flow-cell detection system. Solutions containing 0.10-5.0 Bq/L of 99Tc, as T99cO4- were pumped through a flow cell packed with extractive scintillating resin used in conjunction with a Beta-RAM Model 5 HPLC detector. While T99cO4- accumulated on the resin, time series data were collected. Control chart methods were applied to the data using statistical algorithms developed in MATLAB. SR charts were constructed using Poisson (Poisson SR) and Gaussian (Gaussian SR) probability distributions of count data to estimate the likelihood ratio. Poisson and Gaussian SR charts required less volume of radioactive solution at a fixed concentration to exceed the control limit in most cases than 3-σ and CUSUM control charts, particularly solutions with lower activity. SR is thus the ideal control chart for low-level on-line radiation monitoring. Once the control limit was exceeded, activity concentrations were estimated from the SR control chart using the control chart slope on a semi-logarithmic plot. A linear regression fit was applied to averaged slope data for five activity concentration groupings for Poisson and Gaussian SR control charts. A correlation coefficient (R2) of 0.77 for Poisson SR and 0.90 for Gaussian SR suggest this method will adequately estimate activity concentration for an unknown solution.

KEYWORDS:

; Beta radiation detection; Methyl dioctylamine; Pertechnetate

PMID:
29724457
DOI:
10.1016/j.jenvrad.2018.04.019
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center