Details

Autor(en) / Beteiligte

• Davis, Adam J
• Strom, Daniel J

Titel

Uncertainty and variability in historical time-weighted average exposure data

Ist Teil von

• Health physics (1958), 2008-02, Vol.94 (2), p.145

Ort / Verlag

United States

Erscheinungsjahr

2008

Link zum Volltext

Quelle

MEDLINE

Beschreibungen/Notizen

• Beginning around 1940, private companies began processing of uranium and thorium ore, compounds, and metals for the Manhattan Engineer District and later the U.S. Atomic Energy Commission (AEC). Personnel from the AEC's Health and Safety Laboratory (HASL) visited many of the plants to assess worker exposures to radiation and radioactive materials. They developed a time-and-task approach to estimating "daily weighted average" (DWA) concentrations of airborne uranium, thorium, radon, and radon decay products. While short-term exposures greater than 10(5) dpm m(-3) of uranium and greater than 10(5) pCi L(-1) of radon were observed, DWA concentrations were much lower. The HASL-reported DWA values may be used as inputs for dose reconstruction in support of compensation decisions, but they have no numerical uncertainties associated with them. In this work, Monte Carlo methods are used retrospectively to assess the uncertainty and variability in the DWA values for 63 job titles from five different facilities that processed U, U ore, Th, or 226Ra-222Rn between 1948 and 1955. Most groups of repeated air samples are well described by lognormal distributions. Combining samples associated with different tasks often results in a reduction of the geometric standard deviation (GSD) of the DWA to less than those GSD values typical of individual tasks. Results support the assumption of a GSD value of 5 when information on uncertainty in DWA exposures is unavailable. Blunders involving arithmetic, transposition, and transcription are found in many of the HASL reports. In 5 out of the 63 cases, these mistakes result in overestimates of DWA values by a factor of 2 to 2.5, and in 2 cases DWA values are underestimated by factors of 3 to 10.