introduction.tex

\chapter{Introduction}
\index{Introduction@\emph{Introduction}}
\label{diss_intro}

Nuclear fuel cycles (NFC) are the collection of interconnected processes which 
generate electricity through nuclear power.  
At a minimum, heavy metal resources (U, Th) are removed from the ground
and fissioned in a reactor, releasing energy.  Generally, this energy is converted 
into electricity while the excess process heat is released to the environment.  
After the fuel is burned, it is removed from the reactor and stored as a solid on the surface.
This is known as a the once-through fuel cycle.

This work focuses on methods for assessing an open fuel cycle and a variety of other 
strategies which have been subjected to intensive levels of technology development and 
deployment (TD\&D).

There are many possible fuel cycle strategies that may be implemented in a nuclear power economy.
The ability of nuclear power to recycle its own waste stream affords it distinct advantages.  
Foremost among these is the option to limit the number of deep geologic repositories (DGR) 
that must be built to dispose of waste.   All reactors generate fission products (FP) from 
which further fissioning is not possible.  The vast majority of fuel cycles call for the FP masses 
to be buried.  DGR space is limited and precious, particularly in light of the United States Senate 2009
decision to cease work on the Yucca Mountain Project.  
By closing the fuel cycle it is possible  that only one repository need be built to satisfy 
future conceivable needs.

Repository space is not the sole consideration for NFC designers.  Economic considerations 
are also weighted very heavily.  The total cost of electricity from nuclear power must remain competitive 
with other forms of production.  As all components in the cycle contribute to the overall cost burden, 
different strategies \& technologies used may have disparate levelized electricity costs.

However, increased costs may be deemed acceptable if there is a commensurate value added.
For instance, system designers often seek to improve the implicit resistance a fuel cycle 
has to the proliferation of weapons.  Other considerations include natural resource 
utilization and sustainability, operating capacity, dynamic deployment effects,
embodied energy costs, and political feasibility.  The material balance for a given strategy
affects, and often drives, these cycle wide metrics.

Because NFCs allow for recycling, material balance calculations may require a higher degree of
algorithmic sophistication than is the case for other forms of electricity production.  
Recycle scenarios in which only one or two 
elements from waste streams are re-burned may partially close the fuel cycle.  
Mixed-oxide (MOX) strategies, such as those pursued in France, 
generally recycle only the plutonium stream.  
To compare alternative recycle strategies such as once-through and MOX on the basis of the metrics 
described above, it is important to simulate the nuclear fuel cycle.  
This involves the characterization 
of material flows at each stage in terms of mass, isotopic composition, and time.  This enables 
the coupling of nuclear fuel cycle component design to the design and evaluation of the system as whole.  

Moreover, perturbing a single parameter in the NFC may have global reach over the entire cycle.
For example, in the case of the once-through fuel cycle, altering the initial fresh-fuel 
\nuc{U}{235} enrichment given to a standard light-water reactor changes how much natural 
uranium must be mined earlier in the cycle.  On the back end, changes to the design of 
a reprocessing facility also affect 
how much energy may be extracted from the fuel form and how the waste may be safely disposed.

Due to the high degree of interconnectedness between components 
even in the simplest cycles, the need for a dynamic 
fuel cycle simulator and analysis framework arises.  

Many fuel cycles are studied via pre-defined base case scenarios in which all input parameters
take on static values.  Linear one-dimensional sensitivity studies may be performed which evaluate
the effects of slightly changing a single parameter, while keeping all other inputs
constant.  However, perturbing a single value may push these base case models into a regime in which 
they are no longer valid.  Moreover, important design regions of the parameter space may be 
overlooked by base case analysis due to the inability to perturb several parameters
simultaneously.  
Exploring and analyzing design regimes that are far from pre-set scenarios is necessary for 
correctly designing components based on parameters with global fuel cycle reach.

Strongly coupling component \& cycle design considerations requires that the fuel cycle be 
simulated repeatedly, each time perturbing some aspect of the cycle.  
Decision analysis methods often require many iterations over the design option space
to function in a statistically meaningful way.
Thus it is desirable for
NFC simulations to run quickly.  This in turn requires that component simulation is even faster. 
By capturing only the essential physics in component models, commensurate algorithmic speed 
boosts are obtained.

Essential physics models are fuel cycle component algorithms which remain physically valid in the 
locality on which they are defined.  At a minimum, their inputs are \emph{perturbable} 
within an acceptable range and their outputs respond accordingly.  Such 
models do not seek to compute extraneous parameters that are not of direct importance
to the system at hand.  For example, the flux in the fuel region of a reactor is 
pertinent to the discharge material composition.  However, the neutron current in the 
shielding is not. 
Essential physics models seek reasonable simplifications of more detailed computational 
simulations that preserve physics-driven responses to changes in important system 
inputs.

The work presented herein will develop essential physics models of nuclear power reactors
and incorporate them into a NFC simulation framework.

The first is a fluence-based parameterized reactor burnup model.  
This method seeks to characterize nuclear power
reactors on properties of the material initially loaded into the core.  The nuclides
themselves will be parametrized in terms of the time-evolution of the neutron production 
and destruction rates, their burnup, and their transmutation vectors.  Since no discretization
will be done in energy or solid angle, the model will focus exclusively on computing 
material flows.  Thus this method will have the fewest number of algorithmic steps while 
remaining perturbable. 

Once this reactor model is demonstrated, it will be applied within a fuel
cycle context.  This will show the validity of such a model as a tool for system designers.
Analyzing several strategies, such as a standard once-through fuel cycle, a recyclable uranium
cycle, and a fast burner reactor cycle, will test the dynamic properties of the fluence-based
model.  

Having proved the reactor model inside of various cycles, the NFC framework will be ready to 
perform at scale.  Picking the fast burner cycle from above, as it has the highest degree of complexity, 
many (stochastically chosen) realizations may be performed.  With this hitherto unseen amount of data, 
fundamentally new types of analysis will be needed to parse through the information.  Entropy-based measures
will be considered as a surrogate for traditional linear sensitivity studies.

After considering the abilities and limitations of this multi-scale model, a refinement to 
the original fluence-based reactor model will be proposed.  In many cores, the flux spectrum
evolves along with the composition.  Such effects would not be captured by the previously 
proposed reactor model.  This in turn limits the fuel cycle schema that may be analyzed.
The integration of this multi-energy group model (which 
accounts for internal spectral shifts) into an NFC analysis framework follows
analogously to the work above.

In \S \ref{1g_paper}, the one-energy group reactor model is demonstrated; \S \ref{ses_paper}
uses this essential physics model to simulate a sampling fuel cycles which are perturbations of 
well known base-case cycles.  Next, \S \ref{cts_paper} dramatically
expands the space analyzed by stochastically 
modeling the NFC as a whole. \S \ref{mg_paper} presents a multigroup reactor
model which incorporates spectral changes as a function of burnup.  Finally, concluding remarks
are presented.