Inverse Method Applied to Adaptive Core Simulation

No Thumbnail Available

Date

2002-11-14

Journal Title

Series/Report No.

Journal ISSN

Volume Title

Publisher

Abstract

The work presented in this thesis is a part of an ongoing research project conducted to gain insight into the applicability of inverse methods to developing adaptive simulation capabilities for core physics problems. Adaptive simulation is a simulation that utilizes past and current reactor measurements of reactor observables (e.g. core reactivity and incore instrumentation readings) to adapt the simulation in a meaningful way to improve agreement with reactor observables. To perform such adaption, we utilize a group of mathematical techniques which address the problem of given a current core simulator model and the associated input data (e.g. cross-sections, thermal-hydraulic parameters), how should the values of selected input data be adjusted to improve agreement with observables without changing the core simulator model, (i.e. how can we obtain the best agreement utilizing our current modeling capability). This is usually referred to as an inverse problem, which is difficult to solve due to its ill-posedness nature. Major advances have been made by mathematicians to overcome the ill-posedness nature of such problems. The proposed project is of an exploratory nature serving to develop expertise in this area, to which the nuclear power community has not participated to any great extent over the last two decades since their earlier contribution during the design, research and developments stages of a proto-typical fast breeder reactor. Exploratory research projects, such as this one, serve to develop insight, form general ideas about areas where little expertise is available, and to provide a basis on whether there is potential for the proposed techniques to be useful and successful. The current work addresses BWR core simulators since their prediction accuracy is inferior to PWRs', providing marginally acceptable agreement between measured and predicted core attributes. This implies that BWRs could benefit from utilizing an adaptive simulation tool. In the work done so far, a virtual approach has been utilized in which two versions of a core simulator (i.e. FORMOSA-B) are utilized. The first one represents actual plant data, and is referred to as the 'virtual core'. In that version, the LPRM readings and their associated instrumental noise have been simulated. The second one is an altered version of the same core simulator, in which modeling and input data errors are introduced to give rise to disagreement between the two versions of the core simulator, and is referred to as the 'design basis core simulator'. That disagreement is made to be of the same magnitude as the actual disagreement which exist between plant data and current core simulators in regard to LPRM readings and core criticality. The virtual core observables at nominal conditions, including the noise component of the LPRM readings, are then utilized to adapt the design basis core simulator. A larger set of virtual core observables including those at nominal and various off-nominal core conditions, with and without the noise signal, are then contrasted to those predicted by the 'adapted design basis core simulator'. Results indicate that the disagreement between the adapted design basis core simulator and the virtual core can be decreased by an order of magnitude, indicating the high fidelity and robustness of the adaptive techniques and that adaption can be utilized as an effective noise filter. These favorable results encourage further development of this project. If successful in improving prediction fidelity when utilizing actual plant data as the basis for adaption, this could lead to an increase in design margins and relaxing of technical specifications, which will have a beneficial impact on reactor operation and economy.

Description

Keywords

Parameter Estimation, Regularization, Discrete Inverse Theory, Core Simulators, Adaptive Simulation, ill-posedness, vector spaces optimization.

Citation

Degree

MS

Discipline

Nuclear Engineering

Collections