Run MCNP 5 input file of certain geometry for flux calculation

Join the discussion
Registration is free. Ask a follow-up in this thread, or start your own.
Sign up Log in
8 replies · 3K views
Salman Khan
Messages
36
Reaction score
2
Can any one please explain if I want to run mcnp 5 input file of certain geometry for flux calculation on different surfaces. So far as I know If I increase the NPS (number of particles) it wll give more accurate result but when I increase NPS from 10e9, input file do not run and close within a second ?
 
Last edited by a moderator:
Physics news on Phys.org

Discussion

What error do you get on the command line, what errors are in the output file? What OS and what version of MCNP? What is the input file? What computer is it running on and how much memory does it have?

That is a very high value for NPS and if your result isn't statistically significant there are usually better ways of improving the answer, but I'm surprised it fails to run.
 
Thanks for your comments, I am using laptop 4 Gb ram and 1.7 GHz processer, my input file run and gives result for NPS 1e9 but as I increase NPS to 1e10,
fatal error. entries must be integers appeared in output file
 
Ooohhhh, 1e10 does not work. You have exceeded max int would be my guess.

I would make the problem more efficient. That is an abnormally high value. You could force things with tricks but I would avoid this.
 
Ok got it, but if I have a source of activity let say 1e18, then how the remaining particles wll be compensated ? as we have the possible available option for 1e9 only??
 
Most MCNP runs are time independent because most particle transport is time independent. Two neutrons passing close by do not 'see' one another so one neutron only exists at any one time per thread. Two electrons would affect each other in the real world but this is not simulated.

All the tallies will produce an answer that is per source particle. More source particles simulated will make the answer more trustworthy. The statistics on the numbers will be better as they get closer to the 'true' result.

To simulate 1e18 you could enter nps 1e6, and then multiply the tally at the end by 1e18. The same answer would be true for a source twice as strong, except you would multiply by twice as much. 1e6 might be enough for many problems but too few for some.

The ps in nps does not mean 'per second' btw. I am not sure what it actually stands for but it is the number of particle 'histories' (MCNP speak for seeing what happens to a source particle) to run before it stops. It has nothing to do with the strength of the source.
 
Alex A said:
Most MCNP runs are time independent because most particle transport is time independent. Two neutrons passing close by do not 'see' one another so one neutron only exists at any one time per thread. Two electrons would affect each other in the real world but this is not simulated.

All the tallies will produce an answer that is per source particle. More source particles simulated will make the answer more trustworthy. The statistics on the numbers will be better as they get closer to the 'true' result.

To simulate 1e18 you could enter nps 1e6, and then multiply the tally at the end by 1e18. The same answer would be true for a source twice as strong, except you would multiply by twice as much. 1e6 might be enough for many problems but too few for some.

The ps in nps does not mean 'per second' btw. I am not sure what it actually stands for but it is the number of particle 'histories' (MCNP speak for seeing what happens to a source particle) to run before it stops. It has nothing to do with the strength of the source.
Thanks a lot Alex,
 
Salman Khan said:
Can any one please explain if I want to run mcnp 5 input file of certain geometry for flux calculation on different surfaces. So far as I know If I increase the NPS (number of particles) it wll give more accurate result but when I increase NPS from 10e9, input file do not run and close within a second ?
Just for fun, try writing that as 10000000000 not 1E10. It probably won't make any difference, but it might. I don't recall exactly what the limit on an integer is for MCNP 5, but I seem to recall it was bigger than 1E9.

If you are still getting answers with too large uncertainty, there are a lot of things you can do other than increasing the number of particles. The term you want to search for is variance reduction. The MCNP user manual has quite a bit to say on this. It may be a bit like drinking from the fire hose. If you more questions about this, do come back and ask more. The specific thing you do depends on your exact calculation.