Neutron Multiplicity Counting with the Analysis of Continuous Detector Signals

Doctoral thesis, 2021

Neutron multiplicity counting is a non-destructive assay method for determining the mass of fissile materials (primarily plutonium) using the measured values of the singles, doubles and triples detection rates. Traditionally, the detection rates are obtained from the counting statistics of neutron detectors. The main problem with this approach is that it is sensitive to the overlapping of pulses which, especially at high count rates, lead to dead time losses in the counting electronics. This feature limits the applicability of the method to the measurement of samples with low emission intensities. To overcome this constraint, an alternative version of neutron multiplicity counting has been developed. The new approach is based on the direct analysis of the continuous voltage signals of the detectors (primarily fission chambers). Since the procedure does not rely on counting individual pulses, it is inherently free from dead time losses caused by their overlapping. As a result, the proposed method provides an alternative to traditional multiplicity counting, especially when measuring high intensity samples, like spent nuclear fuel. The thesis presents the complete process of establishing the new version of multiplicity counting. Based on a stochastic model of continuous detector signals, expressions are derived for some of their one- two and three-point (in time) moments (including their mean, covariance function and bicovariance function) and it is shown that the singles, doubles and triples detection rates can be recovered from them. In a computational study, detector signals are simulated and analysed in order to investigate the effect of certain parameters (the measurement time, the detection efficiency, the amplitude of the electronic noise and the intensity of non-neutron pulses) on the estimated values of the detection rates. To demonstrate the practical use of the proposed method, measurements have been performed using a Cf-252 source and the detection rates recovered from the moments of the recorded signals were compared with reference values obtained with pulse counting.