I would recomment GEANT4. It is used to simulate detectors a lot, but it can easily be used to calculate energy loss/deposition and the absorption of essentially any particle (including neutrons). It is used in medicine a great deal for calculation of shields and simulation of things like PET scanners.
https://geant4.web.cern.ch/geant4/
https://en.wikipedia.org/wiki/Geant4
I should add, that to measure the effectiveness of a given shield in the simulation one would put a perfect virtual detector in the location that you wish to be shielded. Then you place neutron sources in the appropriate locations and measure the neutron flux and spectrum through your perfect detector. Then you put your possible shield configurations into place and you measure the flux and spectrum of the neutrons in your virtual detector with the shielding.
Visit the home page of RSICC, Radiation Safety Information Computational Center at the Oak Ridge National Laboratory, http://rsicc.ornl.gov
The header of this page says: "Since its inception in 1962, RSICC has evolved into one of the world’s leading resources for a broad range of the best available nuclear computational tools and services. RSICC software and data collections provide in-depth coverage of radiation transport and safety topics ..."
You can find there less sophisticated codes as well, which help you learning the coupled neutron-gamma transport in shieldings.
If you can use geant4 for your shielding problem or not, and get a meaningful result, depends a lot on your problem. In particular the materials in your setup, the neutron energies involved, and what biasing you need. MCNP is more made for the purpose, but also there you need quite some experience to get things right...
Many thanks for u all i read about Geant 4 But its results has a a large errors. Also MCNP is not free software and RISCC refuse to give me a copy with a discount or free licensed one.
Geant4's errors are really not so large, and one should always be quite quite conservative when designing shielding anyways, so a 5% or 10% error do not mean too much when you are conservative by a factor of 2 or 3.