## R & D PROJECTS

AiNT is very active in the field of research and development in the area of nuclear radiation measurement technology. We develop measuring methods and measurement systems for different applications as well as associated evaluation software and methods. In particular, AiNT is involved in the following publically funded projects:

**ZEBRA (Non-destructive elemental analysis of hazardous substances and contaminated sites)**

Development of an innovative measurement system based on P&DGNAA technology for environmental analysis including new evaluation algorithms in cooperation with the working group "Simulation in Nuclear Engineering" of the Center for Computational Engineering Science at RWTH Aachen University

**PROMETEUS (Process of radioactive Mercury Treatment under EU Safety-standards)**

Development and validation of a disposal concept for radioactive mercury and mercury-containing waste contingents In cooperation with the Institute for Energy and Climate Research - Nuclear Waste Disposal and Nuclear Safety at Forschungszentrum Jülich

#### Development of an innovative Multipeak Analysis Methodology

The fully automatic evaluation of gamma spectra, especially for industrial applications, is a challenge for which whether conventional software or satisfactory off the shelf solutions exists. Therefore, we independently develop novel methods of evaluation. We follow the maxim of using all available information for the evaluation in order to achieve complete results with the lowest possible uncertainties. Our approach is based on the simultaneous evaluation of all peaks of an element in the recorded gamma spectrum. In two steps, a so-called fingerprint of each element is first located within the spectrum in order to identify all elements contained in the sample. Then the main characteristics of the spectrum are mathematically reconstructed to determine a complete composition (w.r.t. mass fractions) of the sample.

This work is being developed in cooperation with the working group "Simulation in Nuclear Engineering" at the Center for Computational Engineering Science of RWTH Aachen University within the research project ZEBRA.

#### Determination of measurement uncertainties according to DIN ISO 11929

Requirements for the determination of measurement uncertainties in nuclear radiation measurement technology are laid down in DIN ISO 11929. These include procedural rules for the calculation of standard applications, a normative definition of important terms (decision limit, detection limit, confidence limits) and, in particular, the requirement for a complete and transparent documentation of both the measurement method and the method for determining the measurement uncertainties.

An uncertainty analysis therefore starts with the definition of the so-called model of measurement. The model can be given by a mathematical formula or by a computer program. Then follows a complete analysis of all parameters which influence the final measurement result. Uncertainties on these variables must be quantified or it is to be documented why uncertainties can be neglected (eg. those on physical constants with very low uncertainties). These individual uncertainties can be quantified by reasoned model assumptions (eg. the underlying statistics of counting measurements), repeated direct measurements or even simulation studies. The propagation of individual uncertainties and thus the uncertainty on the measurement result are determined by means of established calculation rules. A complete documentation of the measurement result then includes the result itself and the associated uncertainty, an indication of all influencing variables, their uncertainties as they were quantified, and the specification of the characteristic limits in a measurement report.

#### Development of a Gamma-Spectroscopy Software

We develop an in-house gamma spectroscopy software for the prompt and delayed gamma neutron activation analysis (P&DGNAA). Hereby, we implement algorithms according to the state of science and technology. We include current and verified nuclear physics databases (eg. IAEA or ENDF-B.VII.1). With these, we can assign and evaluate the peaks in the gamma spectrum. The net peak area is determined by fitting a physical model to the data. Thereby, the masses of each element in a sample can be calculated. The determination of necessary measuring parameters such as energy-dependent photopeak efficiencies or integral neutron flux is currently carried out using Monte-Carlo simulations on our high-performance computing cluster. We are currently working on methods that should lead to a complete automation of spectra evaluation and which enable multi elemental analysis within seconds.

#### Development of advanced uncertainty analysis methods using Bayesian statistics

Standard methods for the determination of measurement uncertainties usually assume complete independence of all input variables and therefore neglect additional (physical) boundary conditions. Bayesian statistics is a powerful mathematical tool that enables a more detailed modelling of uncertainties by adding additional information. We use this tool to further develop the standard procedures for the determination of measurement uncertainties in cooperation with our colleagues from MATHCCES at RWTH Aachen University. This results in a more precise quantification of the uncertainties and usually leads to a reduction of uncertainties. The simplest example of additional information that can be included in the calculation of measurement uncertainties is non-negativity of physical quantities. But Bayesian statistics also allows for more complex considerations, such as a hierarchical evaluation of measurement results. Results from previous steps are included in the evaluation of the current step and thus reduce uncertainties. An example for this is the determination of mass fractions of individual elements in neutron activation analysis. If several masses have already been determined and the total mass of a sample is known, uncertainties in the determination of the masses of other elements contained in the sample are reduced.