ERRORS AND ERROR PROPAGATION IN COMPUTED TRU ACTIVITY
DOE/ERSP TECH NOTE NO, 20.0
DATED: 5 March 1980
AUTHOR: Madaline Barnes, DRI
The data used in computing TRU activity were of several different types, and each type came from
a different source. The bulk of the data was measured values of
441Am provided by EG&G and
extracted from spectra generated by the IMP detector. The peak areas were computed from net
photopeak count rates, and the conversion to pCi/g was made using a factor determined by EG&G,
In some cases correction factors related to the detector were also applied. The determination of
when to apply such corrections and the amount of the correction were made by the EG&G
scientist. The statistician received the uncorrected
24lam values, and the list of corrections, if
necessary. The actual corrections were always made by the statistician to reduce confusion and
error.
Data used for computing TRU to 24!am ratios were provages488
cee p these consisted of data from
aS
chemical and alpha spectroscopic analysis of soil for 2
40Py and a gamma analysis for
lam. Some samples also were analyzed chemically for 2a to provide a check on the gamma
results. The gamma spectra were analyzed using methods very similar to those used by EG&G. The
ratio was computed by the statistician, usually with 241 4m by gamma; sometimes 24. 4m by
ehemistry was,used due to detector problems or when samples had low activity. The decision about
which type of 241m data to use was made by the statistician.
The third type of data used in TRU computations was a correction for signa] attenuation to the IMP
detector due to heavy brush. The correction factor, called the Brush Correction Factor (BCF), was
determined empirically to be about 1.15 in an experiment done early in the cleanup on Island Pearl,
which was supervised by the EG&G scientist.
Details of the experiment and computation of the
BCF are in Tech Notes 1.0 and 1.1. The proportion of the detector view that was covered by brush
at each location was determined subjectively by the IMP technician in the field. The information
was added to the stored spectrum at the time of sampling.
TRU = Am x R x (1-Br) + Am x Rx Brx 1.15 = (Amt+0.15 x Am x Br) xR
where
TRU
=
computed activity of 298py + 239,240py + 241 am
Am
=
measured 241am activity
R
=
computed ratio of 238py + 239,240py + 24lam to 24lam
115
=
factor to correct for attenuation from 100% brush error
Br
=
proportion of detector view covered by brush
Poa
The general formula used for computing TRU is:
and
Possible detector-related corrections were adjustments for crystal effective area or changes in
detector efficiency. During one time period in early 1978, one detector was operated at an
incorrect voltage, and corrections had to be made to this data. For details on the voltage
corrections, see Tech Notes 5, 5.1, and 5.2, Whenever any such corrections were required, they
were made on the 241 4m value, which was then used in the general formula.
ad
Sources of Error
Each type of data was subject to various kinds of error, only some of which were included on the
error propagation computation.
B-20-1