Characterisation of neutron fields: challenges in assessing the directional distribution

The SCK.CEN has carried out neutron field characterisation campaigns at several nuclear reactors. The main goal of these measurement campaigns was to evaluate the performance of different neutron personal dosemeters. To be able to evaluate the performance of neutron personal dosemeters in terms of Hp(10), knowledge of the directional distribution is indispensable. This distribution was estimated by placing several personal dosemeters on all six sides of a slab phantom. The interpretation and conversion of this information into a reliable value for Hp(10) requires great care. The data were analysed using three methods. In the first approach, a linear interpolation was performed on three perpendicular axes. In the other two approaches, an icosahedron was used to model the angle of incidence of the neutrons and a linear interpolation or a Bayesian analysis was performed. This study describes the limitations and advantages of each of these methods and provides recommendations for their use to estimate the personal dose equivalent Hp(10) for neutron dosimetry. Neutron personal dosimetry is complicated by the fact that the neutron dose quantity Hp(10) is strongly energy and angular dependent. Instead of simply assuming that the fluence is unidirectional or that the fluence is isotropic, an attempt was made to estimate the directional distribution of the neutron field using a relatively simple measurement procedure. A number of active and passive personal dosemeters were placed on the six faces of a slab phantom and the results were analysed via different algorithms to obtain partial fluences in several directions of incidence. The results from all calculations in this study show the importance of introducing information about the directional distribution of the neutron fluence for the estimation of the personal dose equivalent Hp(10). The difference between Hp(10) dose estimates carried out using a unidirectional or an isotropic distribution can be of up about a factor of 3 to 4, depending on the energy spectrum of the neutron field, while the actual value of Hp(10) will lie somewhere in between, depending on the directional distribution of the neutron field. For the measurements performed in the two nuclear power plants discussed in this study, the resulting value was mostly closer to the uniform distribution. Furthermore, proper evaluation of the measured data by means of a good algorithm is also of major importance. Algorithm 1, where linear interpolation is only performed in perpendicular axes on the front hemisphere, gives a very crude dose estimation. Algorithms 2 and 3 give similar dose results and they both take into account the front and back hemisphere using 20 discrete angular bins, corresponding to the faces of an icosahedron. While algorithm 2 is easier to apply because the mathematics are simpler, algorithm 3 gives a more detailed dose estimation with uncertainty evaluation and statistical information about the consistency of the model, but it requires good knowledge of Bayesian analyses. (authors)