Benchmarking of the RAPID Eigenvalue Algorithm using the ICSBEP Handbook
Butler, James Michael
MetadataShow full item record
The purpose of this thesis is to examine the accuracy of the RAPID (Real-Time Analysis for Particle Transport and In-situ Detection) eigenvalue algorithm based on a few problems from the ICSBEP (International Criticality Safety Benchmark Evaluation Project) Handbook. RAPID is developed based on the MRT (Multi-Stage Response-Function Transport) methodology and it uses the fission matrix (FM) method for performing eigenvalue calculations. RAPID has already been benchmarked based on several real-world problems including spent fuel pools and casks, and reactor cores. This thesis examines the accuracy of the RAPID eigenvalue algorithm for modeling the physics of problems with unique geometric configurations. Four problems were selected from the ICSBEP Handbook; these problems differ by their unique configurations which can effectively examine the capability of the RAPID code system. For each problem, a reference Serpent Monte Carlo calculation has been performed. Using the same Serpent model in the pRAPID (pre- and post-processing for RAPID) utility code, a series of fixed-source Serpent calculations are performed to determine spatially-dependent FM coefficients. RAPID calculations are performed using these FM coefficients to obtain the axially-dependent, pin-wise fission density distribution and system eigenvalue for each problem. It is demonstrated that the eigenvalues calculated by RAPID and Serpent agree with the experimental data within the given experimental uncertainty. Further, the detailed 3-D pin-wise fission density distribution obtained by RAPID agrees with the reference prediction by Serpent which itself has converged to less than 1% weighted uncertainty. While achieving accurate results, RAPID calculations are significantly faster than the reference Serpent calculations, with a calculation time speed-up of between 4x and 34x demonstrated in this thesis. In addition to examining the accuracy of the RAPID algorithm, this thesis provides useful information on the use of the FM method for simulation of nuclear systems.
General Audience Abstract
In the modeling and simulation of nuclear systems, two parameters are of key importance: the system eigenvalue and the fission distribution. The system eigenvalue, known as kef f , is the ratio of neutron production from fission in the current neutron generation compared with the absorption and leakage of neutrons from the system in the previous neutron generation. When this ratio is equal to one, the system is critical and is a self-sustaining chain reaction. Knowledge of the fission distribution is important in the nuclear power industry, as it enables engineers to determine the best reactor core assembly configuration to maintain an even power distribution. Several methods have been developed over the years to effectively solve for a nuclear systems fission distribution and system eigenvalue. Aspects of both Monte Carlo and deterministic transport methods have been combined into RAPID’s MRT methodology. It is capable of accurately determining the system eigenvalue and fission distribution in real time. This thesis examines the accuracy of the RAPID algorithm using four unique problems from the ICSBEP handbook. These problems help us to test the limits of the FM method in RAPID through the modeling of small, unique geometric configurations not seen in large, uniformly configured power reactor cores and spent fuel pools. For comparison, each problem is modeled using the Serpent Monte Carlo code, an accurate code meant to serve as the industry standard for determination of the fission distribution of each problem. This model is then used to generate a set of FM coefficients for use in RAPID calculations. It is demonstrated that the eigenvalues calculated by RAPID and Serpent agree with the experimental data within the given experimental uncertainty. The fission distribution obtained by RAPID is also in agreement with the Serpent reference model. Finally, the RAPID eigenvalue calculation is significantly faster than the corresponding Serpent reference model, with speed-ups ranging from 4x to 34x demonstrated.
- Masters Theses