Number of bins influence the USRBIN results

Hello FLUKA experts,

I want to simulate an x-ray source of 6MeV. The beam at a distance of 3m from the source is a rectangular of 1.7mX0.002m. (very thin and very wide)

I also know that the dose rate is 8Gy/min at distance 1m from the source. The leakage is 2.0E-5.

I need to calculate the dose rates when I put this source in a room.

I wanted to use USRBIN to get the does rates at 1m so I can normalize it to 8Gy/min. But the results I get from the USRBIN change when I change the number of bins.

When I use a point source or even the same source but with a big square width instead (1mX1m) the change in the bins will change only the resolution of the results (as expected).

My guess is that my bins are not small enough, but in order to increase the number of bins I need to measure a smaller area. Which is a problem, because at the end I need to put the source in a room and measure the dose rates everywhere.

I’m not sure how to proceed. How should I normalize my source?

Attached my files

Many thanks in advance,


XrayBeam.flair (6.5 KB)
XrayBeam.inp (1.5 KB)
BeamSource.f (9.3 KB)
spec.dat (2.0 KB)

Dear Yael,
as you already mentioned USRBIN starts to be sensitive to dimensions or number of points if statistic (number of primaries per scored region) is not sufficient. To resolve this problem I may suggest you two solutions.
The first is easier - you may create USRBIN region accurately to the beam profile dimensions at the area where you want to get normalization data and use it just for the calibration purpose. Make it short along the beam axis, set reasonable resolution and primary number. After you can make other USRBIN regions according to your tasks.

The second, for me more elegant and efficient for larger projects, is to use USRBIN regions according to your tasks, but to use BIASING technique, namely Region Importance Biasing (Surface Splitting). This approach will increase statistics exactly for the region, that you will use for the normalization and where improved statistics is necessary.
There is a good presentation about this technique at the Official FLUKA CERN Web pages

Kind regards,