Nuclear Material Storage Site Selection Using Geo-Cyber Analysis
Undergraduate Students at Texas A&M University
*Authors listed in alphabetical order.
Industrial and Systems Engineering:
Michelle Bermudez bermm08@hotmail.com
Thomas Gray tomegray77@neo.tamu.edu
Dustin Hirner dustinhirner@gmail.com
Abstract
Nuclear Material Storage Site Selection Using Geo-Cyber Analysis will
combine facility location and site selection methods to analyze appropriate
locations for the storage of nuclear materials, including spent fuel and
radioactive waste, under multiple criteria. This project will combine traditional
techniques in facility location (mathematical modeling, network optimization)
with spatial analysis tools and Geographic Information Science (GIS). New
nuclear material storage facility locations must balance the location’s potential
for cyber exposure with its physical (geographic, environmental) vulnerabilities.
Students on this project will work to develop measures for site suitability and
facility vulnerability as well as mathematical models for the location of a single
nuclear materials storage facility and a set of nuclear materials storage
facilities.
Nuclear Engineering:
Christina Kalich c_kalich@tamu.edu
Jacob Landman jakeypoolandman@tamu.edu
Can Pu pucan1991@gmail.com
Our model is based off the p-median problem.3 This type of model is a facility location problem that
locates P facilities in relation to customers so that the shortest distance is chosen between a facility
and repository. It minimizes the product of the weighted Euclidian distance between repositories
and facilities and facility production. As the problem size increases, the time required to solve the
problem exponentially increases (NP hard) which requires a heuristic algorithm to solve it. For our
model it is assumed that high level nuclear repositories are un-capacitated and production of high
level nuclear waste from nuclear facilities is based off of net megawatts of electrical energy
produced by each facility.
Criteria for spent fuel site selection
Relating to:
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Figure 3. Map of the United States: Potential waste site locations using the coordinates of 103 power
reactors and 31 research reactors to minimize the weighted Euclidian distance objective function.
While the model is simplistic, it still produces sites (like Death Valley) that are well supported.\
•
•
•
•
Population Density
Transportation
Climate
Erosion
Dissolution
Tectonics
Human Interference: Natural Resources and
Site Ownership and Control
Surface Characteristics
Host Rock Characteristics
Meteorology
Hydrology
Geohydrology
Geochemistry
Environmental Quality
Socioeconomics
Ease and Cost of Siting, Construction,
Operation, and Closure
Site Ownership and Control
Offsite Installation and Operations
Future Considerations
• Sensitivity Analysis: Determine which
variables in the objective function affect the
final location the most
• Move from proof of concept model to
Matlab Genetic Algorithm that includes
more site selection criteria (seismic activity
and proximity to water)
• Expand the model to include capacity and
production considerations which have
already been collected
• Expand the model from a regional (Texas)
to national scale (United States)
A Genetic algorithm is a heuristic algorithm based on the principle of survival of the fittest.1 For our
model this type of algorithm is useful because it is NP hard. Independent variables are transformed
into binary numbers, where 0 and 1 stand for two different forms of a parent gene (allele). This
algorithm has high global search ability and robustness, and is widely used in non-linear
optimization because of the use of crossover, mutation and selection to maximize the fitness value
of the solution. For our final solution, a MatLab script of this algorithm will be used as opposed to
our proof of concept model made in Microsoft Excel.
.
Figure 4. Using a multiple facility approach decreases the objective function considerably when at
least four locations are used on a national scale.
Figure 1. ArcGIS screen shot of points that represent suitable locations for a nuclear waste
repository. In this figure, no points overlap Native American reservations, aquifers, or counties
consisting of population densities greater than 83 per square mile. Below is a sample of a Python
script to execute ArcGIS tools.
ArcGIS Methods
The first step of the process, which determined which locations would be
suitable for site selection, was to create a fishnet that overlaid Texas. The fishnet
created a grid with points in between each web that contained an x and y
coordinate. Then, counties with high population density and points on Native
American reservations, aquifers, salt domes, and National and State Parks were
deleted off the fishnet grid. Using the near tool, points by rivers were deleted as
well. The remaining points are the suitable locations.
Citations
1.
2.
3.
Acknowledgements
Figure 2. Example of genetic algorithm with the number of generations on the x-axis and the value of the
objective function on the y-axis. After fifteen generations, the objective function begins to plateau. The best
fitness is the largest value, which for this example is 2.70466 with a 0.5% error with respect to the theoretical
value e.
This research was funded by the Aggie Challenge Program at Texas A&M University under the
supervision of Dr. Justin Yates (jtyates@tamu.edu). It would not have be possible without the
assistance of Dr. William Charlton (wcharlton@tamu.edu), the City of College Station GIS
Department, the Nuclear Regulatory Commission, U.S. Census Bureau, U.S. Geological
Survey, ArcGIS Support/Help Menu.****************
4.
Holland, John. Adaption in Natural and
Artificial Systems: An Introductory Analysis
with Applications to Biology, Control, and
Artificial Intelligence. Cambridge: MIT
Press, 1992. Print.
"TIGER/Line Shapefiles." Census Bureau
Homepage. Web. 15 Apr. 2013.
Hakimi, S. “Optimum Locations of
Switching Centers and the Absolute Centers
and Medians of a Graph”, Operations
Research, 12 (1964), 450–59.
"Welcome to the USGS - U.S. Geological
Survey." Welcome to the USGS - U.S.
Geological Survey. Web. 15 Apr. 2013.