MINDS @ UW-Madison

GENIUS v2: An Extensible Platform for Modeling Advanced Global Fuel Cycles

Show full item record


Dunn, Kerry; Ahn, Tae Wook; Elmore, Royal; Huff, Kathryn; Oliver, Kyle; Wilson, Paul P.H.
Nov 04, 2009
network flow; systems analysis; nuclear fuel cycles
GENIUS v2 is a modular extensible platform for the study of advanced nuclear fuel cycles. A great deal of flexibility is available within the high-level modeling approach described here. Fundamental to GENIUS v2 is the modeling of every individual fuel cycle and reactor facility in the nuclear fuel cycle. Each facility is assumed to operate as a black box converting discrete batches of one material, with a given isotopic composition, into discrete batches of new material, with new isotopic compositions. The modular nature of this implementation allows new developers and users to implement any numerical model for achieving this conversion. GENIUS v2 employs an approach in which every facility is owned by an institution, and each institution operates in a specific region. Institutions are intended to represent a generic operating entity, whether a private corporation, a government, an international agency, etc. Regions are intended to represent any type of geo-politically defined part of the globe, real or representative, including sub-national, national, or super-national regions. This ownership hierarchy affords two important capabilities: (a) performance characteristics of facilities can be determined in part by the institution or region in which they operate (b) relationships between/among institutions and regions can be modeled in other parts of the problem. At each time step, a set of requests for material and offers to provide material are collected and arranged into a network flow problem. The sources and sinks for this problem are the facilities that produce/consume the commodity in question. A different network flow problem is created for each commodity in the problem. The behavior of the system is affected by changing the arc costs of the network flow problem. By adopting a modular approach for the definition of the arc costs, users/developers are free to introduce their own behaviors. The default behavior defines default affinities for trade between different institutions and different regions to determine these costs. The user can override these affinities to model different technical/economic/political arrangements. (Affinities are a coarse grained quantity intended to model a qualitative behavior.) A common problem in fuel cycle modeling tools that include advanced fuel cycles is the ability to match the available separated materials with desired fuel recipes. With a discrete material approach, a number of batches of separated material are available at any point in time, each with a specific isotopic composition. We formulate this as a small optimization problem in which we minimize the relative difference between the achieved recipe and the target receipe on an isotope-by-isotope basis while constraining the difference between the achieved and target recipes based on some metric of neutronics performance. Under this paradigm, users/developers are free to modify the objective function and/or constraints to improve the quality of the approximation. The output of GENIUSv2 is a database documenting every transaction of material including: the facility from which the material originated, the facility to which the material was sent, the simulation time step at which the transaction took place, and a reference to the composition of the material at that time. Post-processing tools are provided for extracting and visualizing this data in a variety of ways.
Permanent link
Export to RefWorks 

Licensed under:

Part of

Show full item record

Search and browse


Deposit materials

  1. Register to deposit in MINDS@UW
  2. Need deposit privileges? Contact us.
  3. Already registered? Have deposit privileges? Deposit materials.