RE: Gamma Doserate Modeling

[Shonka Research Associates Inc <sra@crl.com>]

It is hard to diagnose a problem with so little information but my
suggestions are as follows:

1. A point kernel model such as Microshield when used for a large
array that has high density will be very hard to converge
numerically, especially for receivers that are close to the array. 
Try to see if you get consistent results at a great distance, then
assure convergence as you approach the source.  If you cannot attain
convergence, model a smaller piece of the source for near receivers. 
Typically, for a homogeneously mixed source such as Microshield would
model, the last three mean free paths are all that contributes to the
answer, which for compton scattering is about the last 50 grams per
centimeter-squared.  Depending on the source term, convergence for
the final answer will generally be driven by the higher energy
photons.

2. Be aware that you can get "raying" from a regular array of barrels
when a receiver is placed on the line that intercepts the barrel
edges.  This can be approximated by point kernel but should be
addressed by Monte Carlo methods if the desired solution appears to
be near regulatory or compliance levels.

3. Try alternate methods, for example, if you have averaged the
entire array, then the dose rate from a half-infinite object is equal
to the microcuries per gram times the average energy per
disintegration (E-bar).  This may provide an approximation to the
answer.

4. Some versions of Microshield have had difficulty at the lower
energy limit using GP buildup.  Use alternate buildup methods to
assure that you are not encountering numerical problems.

Joe Shonka
Shonka Research Associates, Inc.
sra@crl.com