Nowadays various experimental procedures are there to calculate the thermal conductivity of various materials using various techniques. The most important thing to be considered here is the cost of the experiment to be performed. Here the thermal conductivity calculations are much expensive and time talking and reproducibility is really a matter of questions.
Molecular dynamics simulations are another way, which is emerging to be an alternative to experimental methods, to calculate the thermal conductivity of wide range of materials. Literature exists on many materials. Most of the lterature talks about "Green Kubo" method or "Muller Plathe" method. LAMMPS is a molecular dynamics code freely available for academic institutions in which these algoriths of above mentioned methods are available. Below given links in LAMMPS forum better explains the method and their limitations and usage in LAMMPS software.
Comparision of two molecular dynamics methods:
http://lammps.sandia.gov/threads/msg35408.html
http://lammps.sandia.gov/threads/msg35405.html
http://lammps.sandia.gov/threads/msg35410.html
