skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: A keff Search Capability in MC21

Abstract

The MC21 Monte Carlo code is required to permit an individual geometric component or groups of components to be tagged as ''movable'' within some permissible range. Typical examples of such movable components would be control devices such as translating rods or rotating drums. Given this geometric information, a target multiplication factor (k{sub eff}), and a convergence criterion, MC21 will iterate on movable component positions and return a final position that reflects a k{sub eff} close to the target value. An initial version of this capability is demonstrated through modifications to MC21 that sets the geometry data structures for the movable components, calls the main Fortran-95 solver to compute k{sub eff}, and converges on the final position. This approach uses an adaptive batching algorithm that continually increases the accuracy of each successive MC21 k{sub eff} result as the movable geometry approaches the converged position.

Authors:
Publication Date:
Research Org.:
Knolls Atomic Power Laboratory (KAPL), Niskayuna, NY
Sponsoring Org.:
USDOE
OSTI Identifier:
903082
Report Number(s):
LM-06K143
TRN: US200720%%71
DOE Contract Number:
DE-AC12-00SN39357
Resource Type:
Conference
Resource Relation:
Conference: Joint International Topical Meeting on Mathematics and Computation and Supercomputing in Nuclear Applications, Monterey, CA, April 15 - 19, 2007
Country of Publication:
United States
Language:
English
Subject:
42 ENGINEERING; ACCURACY; ALGORITHMS; CONVERGENCE; GEOMETRY; MODIFICATIONS; MULTIPLICATION FACTORS; TARGETS

Citation Formats

Morrow RE, Trumbull TH, Donovan TJ, Sutton TM. A keff Search Capability in MC21. United States: N. p., 2007. Web.
Morrow RE, Trumbull TH, Donovan TJ, Sutton TM. A keff Search Capability in MC21. United States.
Morrow RE, Trumbull TH, Donovan TJ, Sutton TM. Tue . "A keff Search Capability in MC21". United States. doi:. https://www.osti.gov/servlets/purl/903082.
@article{osti_903082,
title = {A keff Search Capability in MC21},
author = {Morrow RE, Trumbull TH, Donovan TJ, Sutton TM},
abstractNote = {The MC21 Monte Carlo code is required to permit an individual geometric component or groups of components to be tagged as ''movable'' within some permissible range. Typical examples of such movable components would be control devices such as translating rods or rotating drums. Given this geometric information, a target multiplication factor (k{sub eff}), and a convergence criterion, MC21 will iterate on movable component positions and return a final position that reflects a k{sub eff} close to the target value. An initial version of this capability is demonstrated through modifications to MC21 that sets the geometry data structures for the movable components, calls the main Fortran-95 solver to compute k{sub eff}, and converges on the final position. This approach uses an adaptive batching algorithm that continually increases the accuracy of each successive MC21 k{sub eff} result as the movable geometry approaches the converged position.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Tue Jan 09 00:00:00 EST 2007},
month = {Tue Jan 09 00:00:00 EST 2007}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • A description of a robust and flexible movable geometry implementation in the Monte Carlo code MC21 is described along with a search algorithm that can be used in conjunction with the movable geometry capability to perform eigenvalue searches based on the position of some geometric component. The natural use of the combined movement and search capability is searching to critical through variation of control rod (or control drum) position. The movable geometry discussion provides the mathematical framework for moving surfaces in the MC21 combinatorial solid geometry description. A discussion of the interface between the movable geometry system and the usermore » is also described, particularly the ability to create a hierarchy of movable groups. Combined with the hierarchical geometry description in MC21 the movable group framework provides a very powerful system for inline geometry modification. The eigenvalue search algorithm implemented in MC21 is also described. The foundations of this algorithm are a regula falsi search though several considerations are made in an effort to increase the efficiency of the algorithm for use with Monte Carlo. Specifically, criteria are developed to determine after each batch whether the Monte Carlo calculation should be continued, the search iteration can be rejected, or the search iteration has converged. These criteria seek to minimize the amount of time spent per iteration. Results for the regula falsi method are shown, illustrating that the method as implemented is indeed convergent and that the optimizations made ultimately reduce the total computational expense. (authors)« less
  • MC21 is a new Monte Carlo neutron and photon transport code currently under joint development at the Knolls Atomic Power Laboratory and the Bettis Atomic Power Laboratory. MC21 is the Monte Carlo transport kernel of the broader Common Monte Carlo Design Tool (CMCDT), which is also currently under development. The vision for CMCDT is to provide an automated, computer-aided modeling and post-processing environment integrated with a Monte Carlo solver that is optimized for reactor analysis. CMCDT represents a strategy to push the Monte Carlo method beyond its traditional role as a benchmarking tool or ''tool of last resort'' and intomore » a dominant design role. This paper describes various aspects of the code, including the neutron physics and nuclear data treatments, the geometry representation, and the tally and depletion capabilities.« less
  • The geometry kernel of the developmental Monte Carlo transport code MC21 is designed as a combination of the geometry capabilities of several existing Monte Carlo codes. This combination of capabilities is intended to meet efficiently the general requirements associated with in-core design products and, at the same time, be flexible enough to support highly general geometric models. This paper provides a description of the different geometry representations of MC21 and outlines how the geometric data is stored internally through the use of Fortran-90 data structures. Finally, two alternative geometric representations of a published BWR unit assembly model are discussed. Resultsmore » for the two representations are contrasted, including k-effective results, relative memory footprints, and relative computational speeds. While total memory footprint is not noticeably reduced, results show significant speed advantages of one representation. (authors)« less
  • Due to the steadily decreasing cost and wider availability of large scale computing platforms, there is growing interest in the prospects for the use of Monte Carlo for reactor design calculations that are currently performed using few-group diffusion theory or other low-order methods. To facilitate the monitoring of the progress being made toward the goal of practical full-core reactor design calculations using Monte Carlo, a performance benchmark has been developed and made available through the Nuclear Energy Agency. A first analysis of this benchmark using the MC21 Monte Carlo code was reported on in 2010, and several practical difficulties weremore » highlighted. In this paper, a newer version of MC21 that addresses some of these difficulties has been applied to the benchmark. In particular, the confidence-interval-determination method has been improved to eliminate source correlation bias, and a fission-source-weighting method has been implemented to provide a more uniform distribution of statistical uncertainties. In addition, the Forward-Weighted, Consistent-Adjoint-Driven Importance Sampling methodology has been applied to the benchmark problem. Results of several analyses using these methods are presented, as well as results from a very large calculation with statistical uncertainties that approach what is needed for design applications. (authors)« less