Versión de HTML Básico
78
ST 1.2 –
Dosimetría externa
PERSONAL DOSIMETRY STATISTICS AND SPECIFICS OF
LOW DOSE EVALUATION
Avila, Ricardo E.*; Gómez-Salinas, Roberto A., Oyarzún-Cortés, Carlos H.
Comisión Chilena de Energía Nuclear. Chile.
* Responsible author, email: ravila@cchen.cl
The dose statistics of a personal dosimetry service, over 70,000+ readings, display a sharp
peak at low dose (below 0.5 mSv) with skewness to higher values. In spite of the non-normal
distribution, approximately 98% of the doses fall below the average plus 2 standard deviations,
an observation which may prove helpful to radiation protection agencies. Categorizing the
doses by the practice of the exposed worker, that skewness increases with the average value
over the specific practice. A correlation, given by a power function, is observed of the extremity
(hand ring) doses to the equivalent personal dose. The low dose peak of the statistics, above,
has focused our attention on the evaluation of LiF(Mg,Ti) dosimeters exposed at low dose, and
read with Harshaw 5500 readers. The standard linear procedure, via an overall reader
calibration factor, is observed to fail at low dose, in detailed calibrations from 0.02 mSv to 1 Sv.
A significant improvement is achieved by a piecewise polynomials calibration curve. A cubic, at
low dose is matched, at ~2 mSv, in value and first derivative, to a linear dependence at higher
doses. This improvement is particularly noticeable below 2 mSv, where over 70% of the
evaluated dosimeters are found. The use of Harshaw 5500 readers at high gain leads to
frequent values of the glow curve that are judged to be outliers, i.e. values not belonging to the
roughly normal noise over the curve. A statistical criterion is shown for identifying those
anomalous values, and replacing them with the local behavior, as fit by a cubic polynomial. As a
result, thedoses above 0.05 mSv which are affected by more than 2% comprise over 10% of the
data base.