We have developed the 293-group cross-section library SONGLIB with complete reaction pathes, broad-spectrum solving ability and wide range isotopes, which is adopted in the lattice code SONG, the next generation reactor lattice calculation and analysis code. SONGLIB provides the multi-group data for the transport calculation, resonance calculation and burnup calculation of SONG. So the precision and reliability of data is very important. However, the size of the data is very large and there are still a lot of personal factors in determination of the processing parameters and methods. In order to evaluate the library, it’s necessary to carry out the test work comprehensively.
Amount of test work is then undertaken based on three kinds of reactors such as pressurized water reactor with UO2 fuel, sodium cooled fast reactor and new-type molten salt reactor with none-beryllium core. In order to test the adaptivity of the library, 39 cases are designed totally, changing the fuel composition, the absorber compositon, the burnup depth and the working temperature. Only the results of a few cases are given in details to shorten the length of this article. For comparison, the Monte Carlo code MCNP is chosen and the library source of evaluated nuclear data is ensured to be the same. The model that has been calculated is simple, the cell with “fuel-cladding-modulator” the three-layer structure and total reflection boundary for the aim of minimizing the differences between the codes. From the macroscopic angle, the results of infinite multiplication factor fit well with each other in total. From the microscopic angle, the absorption rate for actinides, fission products and absorbers, the fission rate for fissile isotopes and scattering rate for moderators are compared with that of MCNP whose output is normalized to the actual cell power. The difference between the reaction rates for most isotopes can be neglected. But there still exist some obvious differences for some isotopes whose resonance effect should have been considered, such as Fe56, ZrNat and W, which might affect the macoscopic results to some extent.
As a conclusion, the data of SONGLIB can be expected with high precision and reliability. And SONGLIB is adapted in solving various problems for reactors with different spectrum, depth of burnup, operation condition or fuel cycling. Nevertheless, there still exist some factors that may affect the final results during the process and use of the library, which should be paid attention to while taking a further step in optimizing and updating the library.