- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Alberto Fasso <fasso_at_mail.cern.ch>

Date: Fri, 14 Mar 2014 14:21:19 +0100

Dear Georgios,

the volume normalization of USRBIN results has been recently discussed

at length. The situation is clearly explained in Note 13 of the USRBIN

command:

The results from USRBIN are normalised per unit volume and per unit

primary weight, except for region binnings and special user-defined

binnings which are normalised per unit primary weight only,

The reason for this is obvious: Cartesian and R-Z (or R-Z-Phi) bins have

a regular geometrical shape and their volume can be calculated

analytically. Region binnings can have any shape, because regions can

be described in many unpredictable ways. Therefore the output of region

binnings is not normalized to the region volume, or in other words a volume

of 1 cm3 is assumed. It is left to the user to divide by a volume, if

known.

It is presently discussed whether an option should be offered in future to

calculate FLUKA region volumes by Monte Carlo, probably off-line.

Alberto

On Fri, 14 Mar 2014, Georgios Dedes wrote:

*> Dear FLUKAers,
*

*>
*

*> I've been doing some beginner's mental exercise on USRBIN ENERGY and
*

*> DOSE scoring. I came up to some conclusions on how it works, but I would
*

*> like to have some feedback from the experts.
*

*>
*

*> First of all, I am using FLUKA 2011.2b.5 and flair 1.2-4.
*

*>
*

*> The setup is extremely simple. An Al (d=2.6989g/cm^3) cube of size
*

*> 2x2x2cm^3, spanning from -1.0 , 1.0 in all dimensions.
*

*>
*

*> I use a linear source (Deltax=0.5cm) of protons of 1MeV, such us that
*

*> the line source is almost equally splitted between two voxels. The
*

*> protons stop in the voxel that are created and the energy is completely
*

*> contained in those 2 voxels. I run 1000 primaries in 1 cycle, in order
*

*> to keep things simple.
*

*>
*

*> Now the scoring:
*

*>
*

*> I score either in X-Y-Z USRBIN or the whole target region. For the X-Y-Z
*

*> case, the target cube is divided in x,y,z as 3,2,2. So 12 voxels
*

*> dividing a volume of 8cm^3, resulting in a volume per voxel Vi=8/12 cm^3.
*

*>
*

*> I have:
*

*>
*

*> 1. X-Y-Z ENERGY USRBIN
*

*> 2. Region ENERGY USRBIN
*

*> 3. X-Y-Z DOSE USRBIN
*

*> 4. Region DOSE USRBIN
*

*>
*

*> I write both in bin and ASCII, in different logical volumes so as to
*

*> make it idiot proof.
*

*>
*

*> Now my results:
*

*>
*

*> - In the ASCII file of the X-Y-Z ENERGY USRBIN, I get:
*

*> 7.9027E-03 7.0856E-03 0.0000E+00 0.0000E+00 0.0000E+00
*

*> 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
*

*> 0.0000E+00 0.0000E+00
*

*>
*

*> As all the energy is almost equally splitted in two voxels, the
*

*> Energy/Primary/Vi = 5GeV/ 1000primaries / (8/12)cm^3 = 7.5E-3 (the
*

*> disagreement with the above comes because I assume exactly equal energy
*

*> deposit)
*

*> So the results is as expected, which means that the ENERGY in the ASCII
*

*> is energy density [GeV/primary/cm^3], normalized per primary and per
*

*> voxel volume
*

*>
*

*> - In the ASCII file of the region ENERGY USRBIN, I get:
*

*> 1.0000E-02
*

*>
*

*> which mean that the energy deposit per primary is assigned to the whole
*

*> target volume without any volume normalization (it is not the energy
*

*> density anymore) [GeV/primary]. Otherwise it would be 1E-2 / 8cm^3. So
*

*> for the region NO VOLUME normalization is applied. Correct?
*

*>
*

*> - In the ASCII file of the X-Y-Z DOSE USRBIN, I get:
*

*> 2.9280E-03 2.6253E-03 0.0000E+00 0.0000E+00 0.0000E+00
*

*> 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00 0.0000E+00
*

*> 0.0000E+00 0.0000E+00
*

*>
*

*> which is the dose per primary, approximately: 5GeV / 1000primaries /
*

*> [(8/12)cm^3 * 2.6989g/cm^3] = 2.7789 (the disagreement with the above
*

*> comes because I assume exactly equal energy deposit)
*

*> So the dose is calculated per voxel in [GeV/g] and a total dose over the
*

*> whole volume should be the average of all voxel doses.
*

*>
*

*> - In the ASCII file of the region DOSE USRBIN, I get:
*

*> 3.7022E-03
*

*>
*

*> which is obviously not the real dose per primary in the whole Al target
*

*> cube. If this was the case the dose would be either the average of the
*

*> doses in the X-Y-Z DOSE USRBIN ASCII file, or the total energy deposit
*

*> divided by the total mass of the region:
*

*> Edep / Nprimaries / (8cm^3 * 2.6989g/cm^3) = 10GeV / 1000primaries /
*

*> (8cm^3 * 2.6989g/cm^3) = 4.632E-4 Gy/g
*

*>
*

*> So it seems that the dose it assigns to the whole region is the total
*

*> Energy deposit per primary in the region assuming a unit volume (1cm^3).
*

*> Right? Note that changing in the REGION card the volume to 8cm^3, won't
*

*> change the result. So in general the region scoring is not aware of the
*

*> volume, which is OK for the energy, but one should be aware of that when
*

*> scoring dose. Are my assumptions so far correct?
*

*>
*

*> Finally, when I take a look into the 2D and 1D plots from flair, using
*

*> the bin files produced:
*

*>
*

*> - The 2D xy ENERGY plot has as expected only two pixels filled, with
*

*> numbers close to 0.00375. This corresponds to the energy deposit in the
*

*> voxel that it actually happened, but normalized using the whole
*

*> projected volume on the single x-y pixel. That means 2 voxels depth in
*

*> z: 5GeV / 1000primaries / (2*8/12cm^3) = 0.00375 [GeV/primary/cm^3]
*

*>
*

*> - The 1D x ENERGY plot gives again as expected only two bins filled,
*

*> with an average value of 0.00375. This corresponds to a normalization to
*

*> 4 voxels, as 4 y-z combinations will be projected to the same x bin:
*

*> 5GeV / 1000primaries / (4*8/12cm^3) = 0.001875 [GeV/primary/cm^3]
*

*>
*

*> Similarly for the Dose...
*

*>
*

*>
*

*>
*

*>
*

*> Thanks in advance,
*

*> George
*

*>
*

*> --
*

*> Dr. Georgios DEDES
*

*> Ludwig-Maximilians-Universität München (LMU)
*

*> Medical Physics Chair (LS Parodi)
*

*> Am Coulombwall 1
*

*> 85748 Garching
*

*> Tel:+49 (0) 89 289-14022
*

*> Fax:+49 (0) 89 289-14072
*

*>
*

*>
*

Received on Fri Mar 14 2014 - 15:30:21 CET

Date: Fri, 14 Mar 2014 14:21:19 +0100

Dear Georgios,

the volume normalization of USRBIN results has been recently discussed

at length. The situation is clearly explained in Note 13 of the USRBIN

command:

The results from USRBIN are normalised per unit volume and per unit

primary weight, except for region binnings and special user-defined

binnings which are normalised per unit primary weight only,

The reason for this is obvious: Cartesian and R-Z (or R-Z-Phi) bins have

a regular geometrical shape and their volume can be calculated

analytically. Region binnings can have any shape, because regions can

be described in many unpredictable ways. Therefore the output of region

binnings is not normalized to the region volume, or in other words a volume

of 1 cm3 is assumed. It is left to the user to divide by a volume, if

known.

It is presently discussed whether an option should be offered in future to

calculate FLUKA region volumes by Monte Carlo, probably off-line.

Alberto

On Fri, 14 Mar 2014, Georgios Dedes wrote:

Received on Fri Mar 14 2014 - 15:30:21 CET

*
This archive was generated by hypermail 2.3.0
: Fri Mar 14 2014 - 15:30:22 CET
*