Reconstruction of the pin power distribution from the HANARO fuel assembly gamma scanning

To determine the pin power distribution without disassembling, HANARO fuel assemblies were gamma-scanned using the tomography method and then the pin power distribution is reconstructed by using numerical methods. The gamma rays emitted in proportion to the pin power distribution are measured by a NaI(T1) detector rotating an assembly at every 10 degrees. The pin power distribution x={ x₁, x₂, ..., x_m} and the measured counts b are related in a coupled set of equations as Ax=b. The elements in attenuation matrix A, which give the possibility that a photon emitted from each pin will be detected by the detector, are calculated using the MCNP code. The iterative least squares method (LLSM) and the wavelet singular value decomposition method (WSVD) are chosen to solve this problem. An optimal convergence criterion is used to stop the iteration algorithm to overcome the divergence in LLSM. WSVD gives a little better results and is more stable than LLSM. The averaged values from two methods give the best results. Since the measured pin power data are not available yet, the calculation results using MCNP are used as the reference data in this thesis. The RMSE(root mean square error) to the reference data are 5.1, 6.6, 5.0, 6.5, and 6.4% and the maximum errors are 10.2, 13.7, 12.2, 13.6, and 14.3%. Since the reference data calculated by MCNP have a statistical error of 3%, the reconstructed results would be close to the real distribution. This thesis presents that it is possible to reconstruct the pin power distribution from the assembly gamma scanning without disassembling. It is found that the effect by random positions of pins is important. Although the effect can be accommodated by the iterative calculations simulating its random positions, it is recommended to use the experimental equipment without a slit in order to obtain more accurate results