Improvements to the PSBR Fuel Management and Neutronics Model in Support of the New Core Moderator Assembly
Open Access
- Author:
- Bascom, Andrew John
- Graduate Program:
- Nuclear Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- April 19, 2019
- Committee Members:
- William J Walters, Dissertation Advisor/Co-Advisor
Kenan Unlu, Committee Chair/Co-Chair
Azaree Tresong Lintereur, Committee Member
William J Walters, Committee Member
Joshua Alexander Robinson, Outside Member
Brenden Heidrich, Special Member - Keywords:
- PSBR
Breazeale - Abstract:
- Improved software for the modeling of the Penn State Breazeale Reactor (PSBR) has been developed to accommodate newly installed experimental equipment with an impact upon reactivity and in-core power distribution. A new reactor tower, moderator assembly, and neutron beamports were installed at the Penn State Breazeale Reactor (PSBR) during the summer of 2018, which the existing fuel management model was unable to support. This thesis will discuss the changes made to the fuel management and experimental analysis models in use at the PSBR to accommodate the new experimental facilities. In addition to changes made to allow the model to continue to perform its original functions, improvements that allow simulations to be performed faster and with less risk of human error will be discussed. These improvements include a set of variance reduction parameters generated using a custom quadrature set, and a program to automatically shuffle the core loading to achieve a user-specified reactivity change. The PSBR is a TRIGA Mk III research reactor licensed to operate at 1MW thermal. The reactor uses a mixture of both 8.5 and 12 wt% Uranium-Zirconium Hydride fuel with a complex burnup history dating back to 1965. An in-house code, TRIGSIMS, is used to track the burnup of this fuel and predict power distribution and reactivity effects for changes in core loading by coupling the MCNP and SCALE codes. The recently installed core moderator assembly has a complex geometry that cannot be modeled by the TRIGSIMS code. Rather than implementing changes within the TRIGSIMS source code, a script was prepared to implement the necessary geometry definitions directly in the MCNP input generated by TRIGSIMS. This allows the new core moderator assembly to be analyzed independently of fuel burnup calculations. After adding the capability to model the new moderator assembly, tools were prepared to streamline the analysis process of new core loadings. The preparation of input decks, previously a manual and time-consuming process, was automated such that only a single input deck need be prepared for a new core loading pattern. The code then runs all cases and automatically produces a single output report taking the place of dozens of output files which required manual processing. To accelerate the process of simulating experiments using the new beamports, a set of variance reduction parameters have been generated using the ADVANTG code. The low flux within the beamtubes relative to the core region makes analog simulations difficult and time consuming. Extensive testing was performed using both built in and custom quadrature sets in order to achieve the greatest speedup within Monte-Carlo simulations. The variance reduction technique employed utilizes an estimate of the adjoint flux to define weight windows within MCNP that will preferentially transport particles along the beamtube of interest within a simulation without biasing the result. In addition to increasing the speed of simulations by a factor of up to 96, this weight window based variance reduction can be easily applied to new problems with only minimal modifications to inputs. This allows it to be utilized by PSBR staff and graduate students without the need to commit significant amounts of time to learning variance reduction techniques. A new method of designing core loading patterns will be presented to aid in future fuel management efforts. Software to perform semi random shuffles was written and coupled with an existing capability of TRIGSIMS to estimate reactivity changes using the ADMARCH diffusion code. This approach has been shown to be able to identify loading changes capable of increasing excess reactivity of the core while complying with user specified limits upon reactivity changes and fuel element placement. An example of its use in core design and a plan to use this method in the design of future core loadings will be presented. Analysis of multiple core loading patterns using the new software will be presented. The core loading installed at the time this work was started will be shown to be nominally acceptable for use with the new core moderator assembly, but with little margin for uncertainty. Development of a new core loading, Core 58A, for first use with the core moderator assembly will be presented. This core loading was designed to provide a greater margin for uncertainties in model predictions and allow for compliance with PSBR license requirements in the event that the model underestimated core parameters. Based on the agreement of model predictions and measurements for Core 58A, a plan for development of a new core loading will be presented. Through this work, the fuel management software used at the PSBR has been updated to allow for modeling of new experimental facilities and to streamline the process of designing and analyzing new core loading patterns. Using a new method to adapt quadrature sets to the recently installed beamports, adjoint driven weight window variance reduction techniques have been applied to the PSBR model. Extensive testing has demonstrated that this method allows Monte-Carlo simulations of the PSBR to be performed up to 96 times faster without biasing the results. The software and variance reduction parameter libraries produced are expected to remain in use at the PSBR for years to come.