Experimental Design for the INL Sample Collection Operational Test

Amidan, Brett G. ; Piepel, Gregory F. ; Matzke, Brett D. ; Filliben, James J. ; Jones, Barbara

Pacific Northwest National Laboratory (U.S.)

关键词: Decontamination;    Bacillus;    Probabilistic Sampling;    59 Basic Biological Sciences;    Homeland Security;   

DOI  :  10.2172/921259
RP-ID  :  PNNL-17129
RP-ID  :  AC05-76RL01830
RP-ID  :  921259

美国|英语

来源: UNT Digital Library

【 摘 要 】

This document describes the test events and numbers of samples comprising the experimental design that was developed for the contamination, decontamination, and sampling of a building at the Idaho National Laboratory (INL). This study is referred to as the INL Sample Collection Operational Test. Specific objectives were developed to guide the construction of the experimental design. The main objective is to assess the relative abilities of judgmental and probabilistic sampling strategies to detect contamination in individual rooms or on a whole floor of the INL building. A second objective is to assess the use of probabilistic and Bayesian (judgmental + probabilistic) sampling strategies to make clearance statements of the form “X% confidence that at least Y% of a room (or floor of the building) is not contaminated. The experimental design described in this report includes five test events. The test events (i) vary the floor of the building on which the contaminant will be released, (ii) provide for varying or adjusting the concentration of contaminant released to obtain the ideal concentration gradient across a floor of the building, and (iii) investigate overt as well as covert release of contaminants. The ideal contaminant gradient would have high concentrations of contaminant in rooms near the release point, with concentrations decreasing to zero in rooms at the opposite end of the building floor. For each of the five test events, the specified floor of the INL building will be contaminated with BG, a stand-in for Bacillus anthracis. The BG contaminant will be disseminated from a point-release device located in the room specified in the experimental design for each test event. Then judgmental and probabilistic samples will be collected according to the pre-specified sampling plan. Judgmental samples will be selected based on professional judgment and prior information. Probabilistic samples will be selected in sufficient numbers to provide desired confidence for detecting contamination or clearing uncontaminated (or decontaminated) areas. Following sample collection for a given test event, the INL building will be decontaminated using Cl2O gas. For possibly contaminated areas (individual rooms or the whole floor of a building), the numbers of probabilistic samples were chosen to provide 95% confidence of detecting contaminated areas of specified sizes. The numbers of judgmental samples were chosen based on guidance from experts in judgmental sampling. For rooms that may be uncontaminated following a contamination event, or for whole floors after decontamination, the numbers of judgmental and probabilistic samples were chosen using a Bayesian approach that provides for combining judgmental and probabilistic samples to make a clearance statement of the form “95% confidence that at least 99% of the room (or floor) is not contaminated”. The experimental design also provides for making 95%/Y% clearance statements using only probabilistic samples, where Y < 99. For each test event, the numbers of samples were selected for a minimal plan (containing fewer samples) and a preferred plan (containing more samples). The preferred plan is recommended over the minimal plan. The preferred plan specifies a total of 1452 samples, 912 after contamination and 540 after decontamination. The minimal plan specifies a total of 1119 samples, 744 after contamination and 375 after decontamination. If the advantages of the “after decontamination” portion of the preferred plan are judged to be small compared to the “after decontamination” portion of the minimal plan, it is an option to combine the “after contamination” portion of the preferred plan (912 samples) with the “after decontamination” portion of the minimal plan (375 samples). This hybrid plan would involve a total of 1287 samples.