• cross section
• nuclear training reactor
• OpenMC
• pile oscillator
• Serpent
• uncertainty quantification
• validation

Reliable and accurate values of cross sections experimental data are necessary to obtain trustable values of burnup, criticality, power distribution, etc. for the design and operation of nuclear reactors. In the past, the pile oscillator method was used to obtain tabulated cross sections data experimentally. Therefore, at the nuclear research and training reactor AKR-2 at the Technical University of Dresden (TU Dresden) I performed around 120 pile measurements with samples made of absorbing and non-absorbing materials (copper, gold, graphite, hafnium, indium, iridium, and zirconium) to obtain integral cross section data of these materials and to validate the Monte Carlo simplified model of the AKR-2 in OpenMC and Serpent. Different periodic motion profiles (step and ramp) and frequencies were used. Power oscillations due to the periodic reactivity insertion by the samples are analysed in the time and frequency domain, after correcting for self-shielding and power drift at AKR-2. The uncertainty quantification was performed using distribution functions and signal analysis.
During this work, the OpenMC simplified model of AKR-2 was developed and Serpent model was refined. The pile oscillator validation showed deviations around 10-20% between experimental and simulated values of integral cross section data for absorbing materials. Non-absorbing materials showed larger discrepancies due to low signal-to-noise ratio of power oscillations. The validation of the simplified Monte Carlo models in both codes was also performed by different experiments at the facility: critical experiment, control rod calibration, radial flux measurement and reactivity worth curves of beforementioned samples.
Lessons learned in the experiment realisation for different types of samples, in the data signal processing and analysis constitute building blocks in the objective of turning the AKR-2 into a facility where tabulated data such as thermal cross section and resonance integrals are measurable.