RE: RE: biasing in 800 MeV electron accelerator shielding

From: Joachim Vollaire <joachim.vollaire_at_cern.ch>
Date: Mon, 24 Jun 2013 04:57:27 +0000

Dear Yuanjie,

Leading particle biasing will limit the time spent to track particles in the electromagnetic shower part of the simulation. If you are interested at shielding in the forward direction, it can be of interest.
Still for your problem I believe that region based importance biasing should be considered. It indeed increase the CPU time per primary, however the "extra" time spent on tracking is in the regions were you are interested to increase the statistical significance of your results thus overall there is a gain. The "currency" to consider is not the CPU alone but what is called the Figure of Merit which is equal to the square of the variance of your scored quantity times the CPU time...
Greetings
Joachim

-----Original Message-----
From: 毕远杰 [mailto:biyuanjie_at_tsinghua.org.cn]
Sent: 19 June 2013 01:53
To: Joachim Vollaire; fluka-discuss
Subject: Re: RE: biasing in 800 MeV electron accelerator shielding

Dear Joachim,

Your command is very useful. I use symmetrical cylinder as you said. As the region based importance biasing looks increse simulation time, I still find there is leading particle biasing. So should I use both these biasing methods, or which one is prefered?

Many thanks
Best wishes
Yuanjie



> -----原始邮件-----
> 发件人: "Joachim Vollaire" <joachim.vollaire_at_cern.ch>
> 发送时间: 2013年6月19日 星期三
> 收件人: "毕远杰" <biyuanjie_at_tsinghua.org.cn>, fluka-discuss <fluka-discuss_at_fluka.org>
> 抄送:
> 主题: RE: biasing in 800 MeV electron accelerator shielding
>
> Dear Yuanjie
>
> Using the LAM-BIAS card is a first good approach, you do not indicate the parameters you used, but reducing the interaction length for photonuclear interaction by a factor of the order of 0.02 / 0.05 would be appropriate for this type of problem.
>
> Concerning biasing to compensate the absorption of particles, the most straightforward approach is region based importance biasing. You can make you walls as concentric layers (using RPP / cylinders.... depending on your tunnel shape) and play with the importance ratio to keep the population of tracked particles constant through the shielding. A good approach is to consider the attenuation length of neutrons to define the thickness of the layers and importance ratio between adjacent regions. Note that in FLUKA importances can only take values between 0.00001 and 100000.
> If your geometry is too complicated to split you could use the USIMBS routine to obtain a similar results during particle tracking without changing the geometry.
>
> Then, there is one thing to consider, is your problem ~symmetrical (can you approximate your tunnel by a cylinder ?) ? If this is the case then using the cylindrical binning and plotting a 2D r,z plot you will see that the results are satisfying as the scoring volumes get larger as you penetrate the shielding.
>
> Hoping this help
> Best regards
> Joachim
>
> -----Original Message-----
> From: owner-fluka-discuss_at_mi.infn.it [mailto:owner-fluka-discuss_at_mi.infn.it] On Behalf Of ???
> Sent: 18 June 2013 09:58
> To: fluka-discuss
> Subject: biasing in 800 MeV electron accelerator shielding
>
>
> Dear FLUKA user,
>
> I want to simulate the neutron and photon dose rate after 50 cm concrete and 300 cm soil shielding in a 800 MeV electron accelerator. The photonuclear reactions were activated and LAM-BIAS is used. Since it is a deep penetration problem, the biasing should be used. But I am not clear about what kind of biasing is preferred in such a problem. Could someone give me some suggestion?
> Many thanks!
>
> Best wishes
> Yuanjie
>
>
Received on Mon Jun 24 2013 - 09:52:24 CEST