Re: Why is FLUKA slower than MCNPX?

From: Mary Chin <mchin_at_mail.cern.ch>
Date: Wed, 31 Aug 2011 13:09:03 +0200

Dear fluka-discuss,

Just to prevent this thread from hanging in suspension till eternity...

Guoqing is correct. The tally of transmitted electrons would indeed be
zero. Because for electrons and positrons, if the user sets WHAT(1) of
EMFCUT SDUM=PROD-CUT higher than the source energy, FLUKA will set the
transport cutoff equal to the production cutoff. What happens during the
simulation is that every electron will be killed at the same energy and at
the same depth without statistical spread (i.e. there is no point in
doing Monte Carlo, we may as well use a calculator).

The above applies only to electrons and positrons. The overriding of
transport cutoff by FLUKA is an intended design feature, because CSDA
approximation for electrons and positrons makes little physical sense.

For other particles, however, the full stopping power would indeed take
over, as explained by Alberto.

The peak in the first bin/channel persists whether or not we use multiple
or single scattering. If we look at the evolution (by comparing plots
plotted with increasing number of histories) we can see that the first bin
fluctuates wildly, suggesting undersampling. While 1 mm Pb might not
intuitively appear to cause a penetration problem, here we are looking at
low-energy electrons.

The count in the first bin would have populated lower energy bins, eg if
we apply logarithmic bins. How important to get this peak right, I am not
sure, considering the range of 1 keV electrons -- it depends on the
application.

As responded to in the other thread on the same issue ('Problem of Y-90
beta source with FLUKA'), I found FLUKA to be faster than MCNPX. Also, to
point out that when considering (un)restricted stopping powers, the
implicit/explicit simulation of 'hard collision' is not only the Moller
but also the bremsstrahlung.

:)

On Fri, 29 Jul 2011, Alberto Fasso' wrote:

>
>> Thanks a lot for your answer!
>>
>> I think, if I am remember correctly ,the EM-CASCA option already turned off
>> the delta-ray production.
>
> No, you don't remember correctly. "Delta rays", in the case of electrons
> are called Moller scattered electrons. Their production is regulated by
> EMFCUT with SDUM=PROD-CUT, as I explained in my previous mail.
>
>> In my case, if the production cut of electrons is
>> large then the maximum source energy, the electron tally is always zero.
>
> Wrong again. If the production cut of electrons is large than the maximum
> source energy, the incident electron deposits its energy according to the
> full stopping power, instead of the restricted stopping power plus delta ray
> production. The energy loss is the same, but the energy deposition is
> different: concentrated along the electron path instead of partly
> deposited far from the path by the delta rays.
> The electron tally (whatever you mean by that) will still include the
> incident
> electron, although it does not include the delta rays.
>
>> In addition, the very high number in the first channel still bothers me.
>
> Sorry, I don't understand what you mean by "the first channel". Please
> explain.
>
>> When I use the following MULSOPT card, the first channel looks OK, but it
>> needs even more time. How can I eliminate the "wrong" tally?
> MULSOPT 0.0 0.0 0.0 1.0 1.0 99999999.GLOBAL
>
> Again, I don't understand your reference to "the first channel".
> The MULSOPT command you are showing switches on single scattering, and of
> course the run will take much more time.
>
> Alberto
>
>>
>> Thanks again for your help!
>>
>>
>>> -----????????-----
>>> ??????: "Alberto Fasso'" <fasso_at_mail.cern.ch>
>>> ????????: 2011??7??29?? ??????
>>> ??????: fluka-discuss_at_fluka.org
>>> ????: zhangguoqing01_at_tsinghua.org.cn
>>> ????: Re: Why is FLUKA slower than MCNPX?
>>>
>>> If I am not mistaken, MCNPX is based on Class I condensed histories,
>>> while FLUKA uses Class II. That means that MCNPX makes no distinction
>>> between "soft" electron collisions and "hard" ones (i.e. production of
>>> delta rays).
>>> Because FLUKA produces and transports delta rays, of course it takes
>>> more time. And if you score the electron current on the outer surface,
>>> you don't get only the primary electrons, but also the delta rays.
>>>
>>> You can check this by setting the threshold for delta ray production
>>> higher than the source energy. (Use command EMFCUT with SDUM=PROD-CUT
>>> and WHAT(1) larger than the maximum source energy). That will convert
>>> FLUKA into a Class I code that can be compared with MCNPX.
>>>
>>> Alberto
>>>
>>>
>>> On Thu, 28 Jul 2011, zhangguoqing01_at_tsinghua.org.cn wrote:
>>>
>>>>
>>>> Dear FLUKA Experts,
>>>>
>>>> I am now making some intercomparisions of FLUKA and MCNPX. The simulation
>>>> uses a Y-90 electron source with a 1 mm plexiglass + 1 mm lead shielding.
>>>> The EM-CASCA option was used in the simulation. The cut values and
>>>> production-cut are 1 keV for electrons and gammas.
>>>>
>>>>
>>>> It turned out that FLUKA runs much slower than MCNPX. Is this normal or I
>>>> should use some other options?
>>>>
>>>> I have recorded the electron current on the outer surface of the
>>>> shielding. For the tally of the first channel (0 - 11.5 keV), FLUKA's
>>>> result is much higher than MCNPX's. The FLUKA spectrum show a obvious peak
>>>> at the first channel. Does it make sense?
>>>>
>>>>
>>>> Thanks a lot!
>>>>
>>>>
>>>> FLUKA Newbie
>
>
Received on Thu Sep 01 2011 - 10:35:11 CEST

This archive was generated by hypermail 2.2.0 : Thu Sep 01 2011 - 10:35:12 CEST