Identifying Neutron Shielding in Neutron Multiplicity Counting

Special nuclear material accounting and monitoring is of high importance in the global nuclear field. One of the major tools used for that end is the neutron multiplicity counting technique, which allows the assessment of waste samples, mainly Plutonium, by measuring the neutron radiation emitted from it. Since the neutron multiplicity counting technique is a passive one, the measurements can be biased if there are neutron absorbing materials present, such as Cadmium, which can "shield" the neutron radiation emitted from it. This can happen either unintentionally (the sample can contain materials other than Plutonium) or deliberately (for smuggling attempts). Today, the most common analysis method for these kind of measurements is the "Multiplicity" method, which does not account for a potential shielding effect. In order to deal with such effects, an active measurement is required, the most common in use today is the "Add-A-Source" method, which can be costly and time consuming.

In the work presented, we have attempted to formulate a new method which will allow the detection of a potential shielding effect in passive measurements. The presented method is not intended to provide a correction for this effect, but rather, to inform the user if a further assay is required (this can be useful for facilities where hundreds, or thousands, of samples need to be assessed, and actively testing each one for shielding effect can be highly time consuming and expansive).

The presented method is based on the already established SVM method, which involves sampling the first 3 moments of the neutron counting signal - the Mean, Variance and Skewness. The proposed shielding detection analysis is an extension of this, which involves sampling the 4th moment of the detection signal, the Kurtosis, and using it to identify variations in the effective efficiency of the system. The main advantage of the presented method is that it does not require any additional equipment or measurements (other than the standard ones), and can be applied to any standard neutron multiplicity counting system.

The validation of the theory was done in two steps. First, a shielding "simulation" test, where a clean sample of Plutonium was measured and the shielding effect was simulated by randomly deleting a set amount of neutron detection events from the signal, denoted as F ( F=0.2 means that 20% of the detection events were deleted, which is equivalent to reducing the detection efficiency of the system by 20%). This was performed on a set of 4 measurements, using 3 different commercial systems. In this stage, a go / no go test was formulated, which allowed a 100\% success rate for identifying shielding, for F>0.2. Next, a set of 18 measurements was performed where samples were measured clean (for reference) and using "absorbers" of 2 masses (0.5 and 1.6 kg of polyethylene). These measurements also allowed a go / no go criteria to be formulated, resulting in a 100% differentiation between shielded and non-shielded samples, even for a shielding matrix of only 500 grams.