%
%	Note the rotating package (page 6-19)
%
\documentclass[12pt,twoside]{report}
\usepackage{aries}
\usepackage{ariesref}
\usepackage{psfig}
\usepackage{rotating}
\usepackage{cite}
\usepackage{xspace}

\newcommand{\gesim}{\mathrel{\vcenter{\hbox{$\buildrel>\over\sim$}}}}
\newcommand{\lesim}{\mathrel{\vcenter{\hbox{$\buildrel<\over\sim$}}}}
\newcommand{\AI}{\hbox{ARIES-I}\xspace}
\newcommand{\AII}{\hbox{ARIES-II}\xspace}
\newcommand{\AIII}{\hbox{ARIES-III}\xspace}
\newcommand{\AIV}{\hbox{ARIES-IV}\xspace}
\newcommand{\HT}{\hbox{HT-9}\xspace}
\newcommand{\TI}{\hbox{TITAN-I}\xspace}
\newcommand{\TII}{\hbox{TITAN-II}\xspace}
\newcommand{\DD}{\hbox{D--D}\xspace}
\newcommand{\DT}{\hbox{D--T}\xspace}
\newcommand{\degC}{$^\circ$C\xspace}
\newcommand{\SdegC}{$\,^\circ$C\xspace}

\newcommand{\N}{\phantom{-}}
\newcommand{\Z}{\phantom{0}}
\newcommand{\D}{\phantom{.0}}
\newcommand{\Y}{\phantom{$\sim$ }}
\newcommand{\X}{\times}

\widowpenalty=10000
\clubpenalty=10000
\begin{document}

\title{\bf 6. ENGINEERING OPTIONS}
\author{
\autlist{Mark~S.~Tillack}{Charles~G.~Bathke}{Michael~Billone}
\autlist{James~P.~Blanchard}{Leslie~Bromberg}{Thomas J. Dolan}
\autlist{Laila~A.~El-Guebaly}{J. Stephen Herring}{Hesham~Y.~Khater}
\autlist{Farrokh Najmabadi}{Alexei~L.~Radovinsky}{Robert~Schleicher}
\autlist{Dai-Kai~Sze}{Lester~M.~Waganer}{Clement~P.~C.~Wong}
}
\date{}
\maketitle

\begin{titlepage}
{%baselineskip=1.8ex
\tableofcontents
}
\end{titlepage}

\setcounter{chapter}{5}
\setcounter{page}{1}

\chapter{ENGINEERING OPTIONS}

\section{INTRODUCTION}
\label{s-6-1}

Through its successful operation, the U.S.\ Fusion Demo must be sufficiently
convincing that a utility or independent power producer will choose to purchase
one as its next electricity generating plant.  A fusion power plant which is
limited to the use of presently-proven technologies is unlikely to be
sufficiently attractive to a utility unless fuel shortages and regulatory
restrictions are far more crippling to competing energy sources than currently
anticipated.  In that case, the task of choosing an appropriate set of
engineering technologies today involves trade-offs between attractiveness and
technical risk.  In general, more advanced concepts appear more attractive but
entail critical issues which must be addressed through R\&D programs prior to
a commitment to construct the Demo.

The Starlite project has assembled a set of requirements for commercial power 
plants based on projections of customer needs 25-50 years from now.  Based on 
these, a clear and self-consistent definition of a Demo power plant has been 
identified (see Sec.~1).  The Demo requirements provide goals for 
engineering components which are very challenging to meet.  
Safety and environmental requirements severely limit material choices.  
Performance requirements provide strong incentives to operate at 
high coolant outlet temperature.  Reliability requirements may be the most 
difficult to meet.  Design solutions must be simple, incorporate adequate 
performance margins, and be tested fully prior to Demo operation.

The range of design options for fusion power plants is limited, and has been 
examined in numerous studies during the past 25 years.  Previous studies 
have shown that advanced low-activation ferritic steel, vanadium alloy, 
or SiC/SiC composites are the most credible candidates for the primary 
in-vessel structural material.  An assessment of engineering design 
options has been performed using these three materials and the associated 
in-vessel component designs which are compatible with them.

The choice of structural material constrains many aspects of the designs.
These assessments include consideration of all design aspects related to the
choice of structural material, and not only the characteristics of the
structural material itself.  Material choices for the structures, breeder and
multiplier, shield, and coolants are all considered.  Special attention is
given to the first wall, blanket, shield, divertor, magnets, and the support
systems required to operate them.  For each class of designs, their ability to
meet the Demo requirements is summarized and the key issues are assessed.

In principle, the ability to meet the mission requirements is the only
important attribute for determining the attractiveness of a design concept.
However, fusion is still in a developmental stage in which large uncertainties
remain in the performance predictions, particularly for those concepts which
offer the greatest potential for a more attractive product.  In addition, some
conceptual designs are forced to extrapolate from the existing database
or project future improvements in order to obtain performance levels which are
expected to provide a sufficiently attractive product.  Therefore, it is
essential for any comprehensive assessment to include consideration of the
critical issues which, if not resolved successfully, will lead to the concept
failing to meet its mission requirements.  In some cases, the remaining
uncertainties are so large that a self-consistent design may be impossible,
given the basic system parameters which characterize tokamak power plant
operation.  In other cases, the power plant would likely operate, but under
reduced performance levels, reduced lifetime, or uncompetitive cost.

The goal of the Starlite assessments is to identify the design option which
meets the minimum requirements with the lowest uncertainty in fulfilling the
mission of Demo.


\section{FERRITIC STEEL DESIGNS}
\label{s-6-2}

\subsection{Overview}
\label{s-6-2-1}

Unlike vanadium alloys and SiC composites, ferritic steel is compatible with
several different coolants, and can be used effectively with either solid
ceramic breeders or liquid metals.  Steel is by far the most developed material
option.  Intimate knowledge of the limitations of this class of alloys must be
balanced against the uncertainties which exist in the behavior of V-alloys and
SiC/SiC composites.

Four aspects of the top-level mission requirements have been given special
attention in the assessment: maintainability, thermal conversion efficiency and
performance limitations (which impact cost), the trade-off between material
costs and performance, and activation.  These were determined to be most
significant as discriminators between the various design options.

Two key issues were selected for special attention: consequences of the loss of
ductility and ferromagnetism effects.  Ductility under irradiation is viewed as
the most serious unknown.  Ferromagnetism causes important operational
problems, but these were viewed as solvable if an appropriate R\&D program were
carried out.

The results of the assessment indicate that ferritic steel remains a potential
power plant structural material within a certain restricted range of
performance.  The requirements for waste disposal are likely to be met with
low-activation variants of ferritic steel, such that this was not considered to
be a major discriminating factor.  Its ability to meet the minimum requirement
for competitive cost of electricity is one of the most serious concerns.  The
relatively restricted temperature range of operation and relatively low thermal
conductivity limit the fusion core power density and the net plant conversion
efficiency, leading to larger machines with more restrictive power density
limits.  In addition, it is highly unlikely that steel could be used in the
divertor unless the heat flux can be maintained much lower than current
projections.  In that case, an alternate technology must be identified, further
complicating the system.  Safety implications of ferritic steel were not
adequately assessed, and should be given special attention in future studies.

Ferritic steel components have fewer critical issues as compared with other
material candidates.  However, the loss of ductility under irradiation is
potentially a feasibility issue.  Existing irradiation experiments performed in
the temperature range of 250--400\SdegC in mixed-spectrum reactors result in
ductile-to-brittle transition temperature (DBTT) values of about 250\SdegC
under conditions with simultaneous neutron damage and helium generation.
Similar testing in fast reactors at \hbox{$T \geq 365$\SdegC} with neutron
damage, but very little helium production, results in DBTT less than 0\SdegC
for some alloys.  Without fully understanding the role of He in raising the
DBTT, it is difficult to extrapolate these results to a fusion spectrum which
is characterized by about 10~appm-He/dpa at the plasma surface of the first
wall.  What is clear, however, is that special precautions would be required if
ferritic steels embrittle in a fusion reactor as they do in a mixed-spectrum
reactor and the risk of failure may be significantly higher in power plants
which use a brittle structural material.

\subsection{Design Options}
\label{s-6-2-2}

During the time period of 1975--1990, ferritic steels were a leading candidate
for the structural material in U.S.\ fusion power plant conceptual design
studies ({\it e.g.,} Blanket Selection and Comparison Study,
BCSS~\cite{BCSS}).  These alloys were considered attractive materials for use
with solid-breeder ceramics and helium coolants, although they are compatible
as well with water and liquid-metal coolants.  Since that time, U.S.\
conceptual design studies ({\it e.g.,} ARIES~\cite{ARIES1,ARIES4}) have
evolved towards higher temperature, lower activation structural materials such
as vanadium alloys and SiC composites.  This trend has been encouraged by the
programmatic impetus to fully exploit the inherent safety and environmental
potential of fusion, as well as to deliver an economically more competitive
product.

Ferritic steels remain the leading structural material in European and Japanese
conceptual design studies and R\&D programs.  Both groups are committed to
ferritic steels, and they have developed their own special alloys for fusion
applications.  The European Union (E.U.) has developed and tested alloys
MANET-1 and MANET-2, which are similar to the \hbox{Fe-12Cr-1MoVW} commercial
alloy \HT.  Currently they are working on lower-activation higher-performance
versions of these alloys.  Researchers in Japan have developed reduced
activation ferritic steels designated as F82H, JFL-1, JFL-2 and JFL-3.  These
steels contain 7--9~wt.\% Cr, about 2~wt.\% W and small amounts of V and Ta.
In a collaborative research effort among the three parties (U.S., E.U., and
Japan), the International Energy Agency (IEA)-sponsored Ferritic/Martensitic
Steels Working Group has selected the modified F82H steel for large-scale
production and testing~\cite{Klu94}.  The Russian Federation (RF) also will
participate in future IEA-sponsored developmental efforts involving ferritic
steels.

Proposed solid-breeder designs have involved a variety of shapes and geometries
for the ferritic steel structural material.  For example, a leading BCSS design
calls for the solid breeder to be encased in ferritic steel plates with He
cooling on the outside of the plates~\cite{BCSS}.  European concepts have been
proposed for both the breeder-in-tube (BIT) and
breeder-outside-tube (BOT) design concepts~\cite{Dal95}.  In addition to
the solid-breeder concepts, the E.U.\ has proposed both a self-cooled
lead-lithium design within a He-cooled, box-like, ferritic steel structure and
a water-cooled design (with lead-lithium breeder) with the coolant flowing
through ferritic steel tubes within the rectangular channels of the box-like
structure~\cite{Gia95}. For the helium-cooled designs, the maximum allowable
structure temperature is set at 550\SdegC.  For ferritic
steel in contact with liquid lead-lithium, the maximum allowable interface
temperature is about 450\SdegC. No serious design has been developed for
ferritic steel and self-cooled Li.  Within the E.U.\ and Japan, there is a
concern about the safety implications of using lithium coolant, regardless of
the structural material.  Within the U.S., vanadium alloys are considered to be
far superior to ferritic steel as a structural material for use with liquid
lithium coolant.

\subsection{Material Properties}
\label{s-6-2-3}

Traditional ferritic steels considered for use as structural materials in a
fusion reactor first-wall, blanket and divertor fall into two composition
categories: \hbox{Fe-(10--12)Cr-1Mo} and \hbox{Fe-(7--9)Cr-1Mo}, where the
numbers are in wt.\%.  Unlike austenitic stainless steels, they contain a
higher amount of carbon (0.08--0.2~wt.\%).  The excess of carbon induces a
martensitic phase transformation during cooling which, analogous to
cold-working, results in a high strength, low-ductility material.  The material
is then heat treated at 650-760\SdegC for 0.5--2.5 hours to achieve an optimum
combination of strength and ductility.  The resulting structure is referred to
as martensitic/ferritic and is distinguished from the $\gamma$-Fe
microstructures of austenitic steels.

The higher chrome alloys have been modified to include additions of V and W
({\it e.g.,} the commercial \HT is a \hbox{Fe-12Cr-1MoVW} higher strength
alloy).  The lower chrome alloys have been modified by additions of V and Nb
({\it e.g.,} \hbox{Fe-9Cr-1MoVNb}) to increase the strength of these alloys.
Because of the concerns regarding radioactivity of the alloying elements Mo, Nb
and Ni, reduced-activation versions of the traditional and modified ferritic
steels have been developed for fusion applications.  In particular, W, V, and
Ta are substituted for the Mo and the Nb, while Mn is substituted for most of
the Ni.  As a class of materials, these will be referred to as
\hbox{Fe-9Cr-2WVTa}, where the 9Cr is a generic representation of alloys with
7--9~wt.\% Cr and 2WVTa refers to 2~wt.\% W with small amounts of V and Ta.

Several aspects of the basic material behavior of these alloys are summarized
below.  The temperature range of operation is very important in determining the
thermal and mechanical performance of in-vessel components.  Ferritic steel is
limited in both the upper and lower allowable temperatures, as described in
Sec.~\ref{s-6-2-3-1}.  Neutron irradiation affects the basic material
properties, thus restricting the service life of in-vessel components.  The
existing database relating to irradiation effects is summarized in
Sec.~\ref{s-6-2-3-2}.  Sec.~\ref{s-6-2-3-3} describes some of the advantages of
the use of steel due to well-developed industrial capabilities, and
Sec.~\ref{s-6-2-3-4} reviews some of its attractive nuclear characteristics.

Based on the performance of these materials, design temperature windows of
250--550\SdegC for use with a He coolant and 285--450\SdegC (interface) for use
with Pb-Li coolant have been recommended.  The lower temperature limit is based
on a ductile-to-brittle transition temperature (DBTT) for irradiated ferritic
steel of 250\SdegC\ (except with PbLi, where the melting point must be
respected). 

\subsubsection{Temperature limits}
\label{s-6-2-3-1}

The temperature window for ferritic steels is limited on both the upper and
lower ends.  Based on the data available to date for \hbox{Fe-12Cr-1MoVW} and
\hbox{Fe-9Cr-1MoVNb} alloys, the lower temperature limit would have to be set
at about 250\SdegC for ferritic steels.  This assumes that the design allowable
stresses account for the reduced ductility in this regime and that there is an
adequate design margin for stresses induced during shut-down and during
off-normal events ({\it e.g.,} seismic) while the reactor is at temperatures
below 250\SdegC.  

However, there still would be concern about the behavior of the embrittled
material during shut-down events.  The embrittlement can be partially reversed
by periodic annealing of the structural components at 400--500\SdegC during
shut-down.  This may or may not be a practical solution.  It is hoped that
continued development and testing of the reduced-activation \hbox{Fe-9Cr-2WVTa}
steels will result in a decrease in the minimum allowable operation
temperature.  This remains to be demonstrated for fusion-relevant neutron
damage and He levels.

If the reduced-activation ferritic steels [\hbox{Fe-(7--9)Cr-2WVTa}] are
demonstrated to have performance superior to the above alloys up to 100~dpa
with 100--1000~appm He and down to temperatures lower than the current data
base for these steels (365\SdegC), then the minimum operating temperature can
be lowered.  Also, in future work, the minimum operating temperature should be
based on fracture toughness data (which is not available) rather than DBTT and
on the more sophisticated ASME rules.  While Charpy DBTT data are adequate for
screening purposes, they are not adequate for design analyses.

\begin{figure}[p]
\centerline{\psfig{file=sti-6-2-01.ps,height=3.25truein}}
\caption[f-6-2-01.ps]
{Ultimate tensile strength (UTS) and maximum
allowable primary stress ($S_m$) for unirradiated steels: \hbox{Fe-12Cr-1MoVW }
({\it e.g.,} \HT), \hbox{Fe-9Cr-1MoVNb}, and annealed 316L(N).}
\label{f-6-2-01}
\centerline{\psfig{file=sti-6-2-02.ps,height=3.25truein}}
\caption[f-6-2-02.ps]
{The change in the ultimate tensile strength (UTSi) of irradiated
\hbox{Fe-12Cr-1MoVW} and \hbox{Fe-9Cr-1MoVNb} ferritic steels relative to the
unirradiated ultimate tensile strength (UTSo).}
\label{f-6-2-02}
\end{figure}

The upper temperature limit of 550\SdegC commonly used in E.U.\ and Japanese
design studies with He coolant is well-established, based on the decreasing
tensile and creep strength of ferritic steels with increasing temperature.
\hbox{Fig.~\ref{f-6-2-01}} shows the ultimate tensile strength, along with the
allowable design stress, for \hbox{Fe-12Cr-1MoVW} and \hbox{Fe-9Cr-1MoVNb}, as
compared to values for the E.U.\ 316L(N) steel.  Irradiation produces a
hardening or increase in this allowable stress for temperatures below
450\SdegC, but a softening above about 550\SdegC (see
\hbox{Fig.~\ref{f-6-2-02}}).  It is unlikely that advances made in optimizing
the chemistry and heat-treatment of the reduced-activation alloys will result
in a substantial improvement in this upper temperature limit.  However, the
upper temperature limit is design-dependent.  Designs with lower operational
stresses can tolerate higher temperatures.

In the case of liquid metal coolants and/or breeders, compatibility concerns
would limit the maximum interface temperature to below 550\SdegC.  Values near
450\SdegC are commonly used for ferritic steel in contact with liquid Pb-Li.
Higher temperatures would be allowed for a ferritic-steel/Li interface, but no
designs for this combination of materials have been developed.  However, as
with vanadium alloys, ferritic steels would need an insulating coating to
minimize the MHD pressure drop.  If such a coating were to be developed for
ferritic steels, then the compatibility of the
ferritic-steel/coating/liquid-metal system would determine if there would be a
limit on the interface temperature.

The 550\SdegC upper temperature limit results in gross thermal efficiencies for
the blanket region of about 37\% for a direct He Brayton cycle and about 45\%
for a supercritical Rankine steam cycle.  It is unlikely that improvements in
the reduced-activation alloys would change the upper temperature limit
significantly.  Also, given the anticipated divertor surface heat loads,
ferritic steels would not be practical for the structural material in this
component.  This represents about 20\% of the thermal power of the reactor.
Either an acceptable higher-performance material must be found, or the overall
plant efficiency would be limited to less than 80\% of the blanket efficiency.

\subsubsection{Irradiation database and R\&D needs}
\label{s-6-2-3-2}

Problems associated with the traditional and modified stainless steels are
being addressed within the IEA-sponsored Working Group on Ferritic/Martensitic
Steels, as well as within the three participating parties.  The problems being
addressed are: activation associated with Mo, Nb, and Ni; low tensile and
impact ductility for irradiation below about 400\SdegC; and, to a lesser
extent, loss of tensile and creep strength, particularly at temperatures above
550\SdegC.  Optimization of the chemical composition and the thermo-mechanical
heat treatment of this class of alloys is being performed in an attempt to
develop better ferritic steels for fusion applications.  The focus of the IEA
effort is the modified F82H steel (\hbox{Fe-8Cr-2WVTa}), which has lower
activation as compared to other alloys within this class which contain Mo and
Nb.

An extensive study of the irradiation effects on \hbox{Fe-12Cr-1MoVW} and
\hbox{Fe-9Cr-MoVNb} was conducted under the fission reactor programs.  The
performance of these alloys is quite good in the fast-fission-reactor
temperature range of 400--600\SdegC, particularly when tested in a fast-fission
neutron environment.  For example, the increase in
ductile-to-brittle-transition temperature (DBTT) tends to saturate at about 10
dpa to maximum values of 50--150\SdegC when irradiated at 390\SdegC, with the
\hbox{Fe-9Cr-1MoVNb} showing the lower increases.  However, when these same
alloys are irradiated in mixed-spectrum reactors, with and without extra Ni
added to simulate fusion-relevant He/dpa ratios, at temperatures in the range
of 300--400\SdegC, increases in DBTT as high as 270\SdegC have been observed.
It is still debated within the materials community whether the ductility loss
in the thermal- and mixed-spectrum reactor radiations is due to helium
embrittlement, helium-assisted precipitation-induced embrittlement, and/or
defect and precipitation hardening~\cite{Klu95}.  While the consequences for
fusion applications greatly depend on the hardening mechanisms (first-wall
ferritic-steels will experience helium generation on the order of 10~appm/dpa)
it is not clear how to resolve this issue without a fusion neutron source.

Preliminary results on a reduced-activation \hbox{Fe-9Cr-2WVTa} alloy indicate
DBTT values well below room temperature up to 30~dpa when irradiated in a
fast-fission reactor at T $\geq$ 365\SdegC~\cite{Klu94a}.  It remains to be
demonstrated that the DBTT remains sufficiently low up to higher fluences ({\it
e.g.,} 100~dpa), with up to 10 appm/dpa He generation, in the temperature range
of 250--400\SdegC.  In the near-term, this may require testing of the new
reduced-activation alloys in thermal- and mixed-spectrum reactors such as HFR
and HFIR, concurrent with development of techniques to determine more directly
the causes of the embrittlement previously observed after irradiation in these
reactors.

While some ferritic steels ({\it e.g.,} \hbox{Fe-9Cr-1MoVNb} and
\hbox{Fe-12Cr-1MoVW}) are code-qualified, the new steels being developed may
need to be further characterized to achieve the same status.  Once an alloy has
been identified as promising from a performance perspective, it is necessary to
generate a data base for use in determining design and allowable stresses.  At
a very minimum, the following data are needed:

\begin{enumerate}

\item
A large number of data points for yield strength, ultimate tensile strength and
ductility of the unirradiated material at room temperature for different
grades, lots and heats of this material;

\item
Enough data points at higher temperatures to establish a temperature trend
curve; and

\item
Enough data points for irradiated material at these same temperatures to
establish a hardening and softening trend curve.

\end{enumerate}

These data would be sufficient to establish limits on primary membrane
stresses, primary bending stresses and secondary thermal stresses during
short-time steady-state ({\it i.e.}, non-cyclic) operation.  Fracture
toughness data would be needed to establish limits on peak stresses at
geometric discontinuities and cracks.  Cyclic operation and/or high temperature
operation would require additional data on creep strength, fatigue strength and
creep-fatigue interaction.  If, as expected, the mechanical properties of the
new ferritic steels fall within the scatter bands of the data for the
code-qualified ferritic steels, then no additional effort is needed in this
area.

\subsubsection{Manufacturing capabilities}
\label{s-6-2-3-3}

Ferritic steel has advantages over advanced structural materials ({\it e.g.,}
vanadium alloys) due to its large installed industrial manufacturing and
fabrication capability both in the United States and abroad.  Large nuclear
grade components are currently being manufactured with ferritic steel.  There
is capability to manufacture all types of commonly fabricated shapes including
plates, sheets, wire, and forgings.  All the standard fabrication processes are
well documented, including cutting, forming, welding, forging, and heat
treating.  The unirradiated physical properties are well known and suitably
documented.

Even if the blanket and the plasma-facing components are to be made from
alternative structural materials, parts of the shielding and ex-reactor
components are likely to be constructed of ferritic steel in order to save on
the cost.  In this case, additional problems may arise due to the transition
between different material systems, involving changes in mechanical properties
and corrosion of dissimilar materials.  These problems would be eliminated in a
system using a single material.

Ferritic steels are more difficult to weld than austenitic stainless steels
because of the care needed before and after welding.  Typically, a pre-weld
treatment is needed to increase toughness by heating to $\lesim400$\SdegC for a
sufficient time.  A post-weld heat treatment is needed to temper the welded
joint by heating to 760\SdegC for 2.5 hours and then cooling at a controlled
rate.  While all forms of welding can be used with ferritic steels, ion-beam
and laser-beam welding in an inert atmosphere are preferred because they limit
the heat-affected zone and the degree of chemical interaction.

The reduced-activation modifications to ferritic steels for fusion application
are anticipated to have minimal effect on manufacturing capabilities or
processes.  Thermophysical properties are not expected to change in any
significant way.  The mechanical properties may be affected to some degree, but
not necessarily in a detrimental way.  However, once a developmental alloy has
been selected, it is important to demonstrate the fabricability and weldability
of such an alloy.

\subsubsection{Nuclear characteristics}
\label{s-6-2-3-4}

Steels in general possess attractive nuclear characteristics compared to other
candidate structural materials.  For instance, steel has the best shielding
capability after tungsten, and is superior to V and SiC in attenuating
radiation.  This property is important for the economics of fusion power plants
as it helps reduce the overall size and cost of the machine.  A savings of
10--20~cm in the shield is achievable by employing steel instead of V or SiC as
the shielding material (Sec.~\ref{s-6-2-3-4}).  Another attractive feature of
steel is the ability to multiply the neutron energy.  The energy multiplication
of steel is significantly larger than that of SiC and comparable to that of V.
This feature should enhance the power balance and, when combined with the
superior shielding properties of the steel, will improve the economics of the
fusion power plant.

\begin{table}[b]
\caption{}
\centering{\bf Cumulative Damage Level for the First Wall\\
at the End-of-Life of the First Wall and Blanket}
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l @{\hspace{0.5cm}} c c c}\hline
             &  ferritic steel  &   V     &  SiC  \\
\hline \\
Damage rate (dpa)     &  150    &  150    &  160  \\
He Production (appm)  &  1,980  &  \D980  &  43,030  \\
H  Production (appm)  &  6,640  &  4,670  &  16,090  \\
\\ \hline
\end{tabular}
\label{t-6-2-1}
\end{table}

The structural materials used in the first wall and blanket of fusion power
plants will require frequent replacement due to radiation damage
considerations.  The cumulative damage levels in the structure at the projected
first wall and blanket end-of-life are compared for ferritic steel, V, and SiC
in \hbox{Table~\ref{t-6-2-1}} for a typical 5~MW/m$^2$ neutron wall loading.
It is assumed that the first wall and blanket will be replaced upon reaching
150~dpa (displacements per atom) in both ferritic steel and V, and 3\% burn-up
in SiC.  The dpa limits for ferritic steel and V are based primarily on
extrapolations of swelling and irradiation creep data.  It remains to be
demonstrated that the mechanical properties are adequate up to such high
exposure levels.  Based on the assumptions in \hbox{Table~\ref{t-6-2-1}}, the
ferritic steel, V, and SiC structures of the first wall and blanket should be
replaced after 2.4, 2.4, and 2.6 full power years of operation, respectively.
As indicated, V generates the lowest amount of He and H gases while SiC has a
relatively high gas production level which could deteriorate the
thermo-mechanical properties of the composites.  Note that the He/dpa ratios in
the first wall structures of fusion power plants are 13, 7, and 270 for
ferritic steel, V, and SiC, respectively.

\subsection{Ability to Meet Requirements}
\label{s-6-2-4}

The choice of primary in-vessel structural material impacts the ability of the
device to meet most, if not all, of the top-level requirements.  These include
safety, environmental and economic attributes.  Our examination focused on a
limited set of these, with particular attention to those which may serve to
discriminate between different material choices.  We concentrated on
(1)~attributes which impact the ability of the plant to meet economic goals,
including maintainability, thermal conversion efficiency and economic tradeoffs
between improved performance and the cost of fabricated structures, and
(2)~activation attributes which impact the ability of the device to meet the
waste-disposal and no-evacuation requirements.

\subsubsection{Maintainability}
\label{s-6-2-4-1}

Maintenance of the ferritic steel structures requires special procedures,
particularly if rewelding of components is required.  The irradiated ferritic
structural material will have both radiation damage and transmuted helium
within the base material.  Upon rewelding, these helium atoms migrate to the
grain boundaries and cause problems of poor reweld performance.  Special
processes must be developed to reliably weld irradiated materials.  Post-weld
heat treatment will be required for re-welded ferritic steels.  (The same may
also prove true for vanadium alloys, but the data base is too limited to allow
a conclusion at this time.)  If the ductile-to-brittle transition temperature
is too high, the component to be maintained may be in a brittle condition or
the handling temperature must be raised.  If hydrogen from the plasma has
permeated the material, hydrogen embrittlement may cause handling or
life-limiting problems.  Annealing may solve or minimize some of the above
problems, but this imposes another step in the maintenance process.

The activation level of the structural materials within the plasma chamber also
influences the design and lifetime of the chamber and hot cell maintenance
equipment.  They have to be designed to handle the background radiation levels
inside the chamber and the hot cell.  A ferritic steel which has been optimized
for low-activation and a low DBTT can probably be acceptably maintained.

\subsubsection{Thermal conversion efficiency}
\label{s-6-2-4-2}

Because of large capital cost, it is essential for fusion plants to operate
with relatively higher thermal efficiency in order to achieve a competitive
cost of electricity. Ferritic steel has been examined with respect to its
effectiveness as a structural material for power conversion systems and the
heat transport systems that form an integral part of the power conversion
process.  The heat transport systems include the first wall and blanket, which
transport roughly 80\% of released fusion energy as heat, and the divertor,
which transports most of the remaining energy.  

Ferritic steels are compatible with helium coolant in all respects.  There is
no corrosion concern nor is there a concern with helium ingress causing
embrittlement.  Likewise, ferritic steels are compatible with water/steam,
provided water chemistry is controlled in a manner typical for steam power
plants.  There is a small concern over the compatibility of ferritic steel with
lithium at higher temperatures.  Ref.~\cite{BCSS} suggests a maximum interface
temperature between 535 and 550\SdegC for lithium velocities between 1.5 and
0.5~m/s in a ferritic steel circuit.

The temperature limit (550\SdegC for use with a plasma facing component)
presents a significant concern with respect to heat transfer, especially for
very high heat flux conditions, as is the case for the divertor.  For both the
first wall and the divertor, the coolant must be cold enough to provide a
sufficient temperature difference ($\Delta T$) to enable the heat flux to be
transferred to the coolant without exceeding the ferritic steel temperature
limit on the plasma facing surface.  The $\Delta T$ is composed of two parts,
the $\Delta T$ through the ferritic steel wall (assumed to be 2 mm thick) and
the film $\Delta T$.  \hbox{Table~\ref{t-6-2-2}} summarizes $\Delta T$ values
for plasma-facing components made of ferritic steels and various coolants.
\hbox{Table~\ref{t-6-2-2}} shows that the total $\delta T$'s are so large for
the peak heat flux condition on the divertor that the heat transfer is either
not physically possible or practical for any of the heat transport fluids.
Therefore, ferritic steel is not suitable as a candidate material for the
divertor, unless the peak heat flux is significantly reduced.

\begin{table}[tb]
\caption{}
\centering{\bf Temperature Drop across In-Vessel Components\\
(Ferritic Steel Structure$^{\mathbf (a)}$ and Various Coolants)}
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l @{\hspace{1.5cm}}  c c c c c c c c }
\hline
  &   \multicolumn{2}{c}{Helium}  &&  \multicolumn{2}{c}{Water}
  &&  \multicolumn{2}{c}{Lithium}  \\ \cline{2-3} \cline{5-6} \cline{8-9}
  &  Peak  &  Avg.  &&  Peak  &  Avg.  &&  Peak  &  Avg.  \\
\hline \\
\quad {\bf Divertor:$^{(b)}$ }  \\
Wall $\Delta$T (\degC)   &  391 & \Z79 &&  391  & \Z79 && $(c)$ & \Z79 \\
Film $\Delta$T (\degC)   &  160 & \Z59 && $(c)$ & \Z76 &&  190  & \Z34 \\
Total $\Delta$T (\degC)  &  551 & 138  && $(c)$ &  155 && $(c)$ &  113 \\
\quad {\bf First Wall:$^{(d)}$ }  \\
Wall $\Delta$T (\degC)   & \Z39 & \Z28 && \Z39  & \Z28 && \Z39  & \Z28 \\
Film $\Delta$T (\degC)   & \Z32 & \Z23 && \Z38  & \Z31 && \Z17  & \Z12 \\
Total $\Delta$T (\degC)  & \Z71 & \Z51 && \Z77  & \Z59 && \Z56  & \Z40 \\
\\ \hline
\multicolumn{9}{l}{$(a)$ For a 2-mm thick structure.} \\
\multicolumn{9}{l}{$(b)$ For a peak and average heat flux of 5.0 and
1.0~MW/m$^2$, respectively.} \\
\multicolumn{9}{l}{$(c)$ Not physically possible because $\Delta$T is so large
as to require the coolant} \\
\multicolumn{9}{l}{\phantom{$(c)$} to be in solid phase.} \\
\multicolumn{9}{l}{$(d)$ For a peak and average heat flux of 0.5 and
0.3~MW/m$^2$, respectively.}
\end{tabular}
\label{t-6-2-2}
\end{table}

Considering maximum structure and interface temperatures, ferritic steel can be
used for the first wall with practical values of bulk coolant outlet
temperature limited to 490--500\SdegC.  At this temperature a direct helium
Brayton cycle would have a maximum gross efficiency of 37\%--38\%.  A direct
supercritical-steam Rankine cycle could achieve a gross efficiency of nearly
45\% and a lithium cooled first wall with an intermediate loop and a
supercritical Rankine cycle could achieve a maximum gross efficiency of
42\%--43\%.  In the latter case, an intermediate loop is required because of
tritium generation in the first wall lithium heat transport circuit and
compatibility concerns between lithium and water.

Although the efficiencies are respectable for first-wall heat-to-power
conversion, this represents only 80\% of the total fusion power.  Another
material ({\it e.g.,} vanadium) would have to be used for divertor
heat-to-power conversion in order to convert the divertor heat loads
efficiently.  The impact of the thermal conversion efficiency on the cost of
electricity is explored in the next section.

\subsubsection{Cost assessment}
\label{s-6-2-4-3}

In-vessel components constitute a significant fraction of the total cost of a
magnetic fusion power plant ($\sim$10\%). Therefore, it is desirable to employ
inexpensive structural materials.  The achievable cost of fabricated structures
is still a subject of debate. The complexity of the structures, impurity
levels, tolerances and inspection quality can have a large impact. In this
section, the trade-offs between ferritic steel and vanadium-based power plants
are examined to help understand under which conditions advanced materials offer
significant improvements to the cost of electricity.

Vanadium alloys are relatively new structural alloys. To date, there have not
been major industrial applications to support their development. The three
basic costing elements for a fusion structural material (production cost of
different product forms, fabrication cost, and installation) are still to be
developed for V-alloy. Therefore, there is no statistical information available
to project a reliable unit cost. Only a rough estimate on the lower and upper
bounds of the unit cost can be made. Based on the application of advanced
structural materials for fusion devices, the product form cost will amount to
about 30\% of the unit cost of the final component, including labor. Taking the
1995 cost of advanced Ni-based structural materials as a starting point, the
product form cost is about \$100/kg. If this is assumed to be the same for
V-alloy, then the lower bound unit cost can be assumed to be about \$300/kg.  A
recently placed DIII-D contract for the \hbox{V-4Cr-4Ti} alloy material
necessary for the upper divertor in the Radiative Divertor Program (RDP) shows
a finished V-alloy component cost of about \$600/kg. For a power plant, this
unit cost is expected to be lower due to increased experience far into the
future.

Our analysis is based on the unit cost of the V-alloy structure at \$300/kg,
ferritic steel structure (MHT-9) at \$68/kg, and Tenelon filler at \$25/kg in
constant 1992 dollars.  It should be noted that the cost of ferritic steel
structures is also uncertain as large-scale industrial infrastructure for these
material does not exist.

The assessment was performed on a ``systems level'' using the ARIES System Code
(ASC). To provide a common basis for comparison, the \AII design concept was
adopted.  \AII is a second-stability tokamak with a radiative divertor and
``close-in'' cryostat (wrapping around the outboard TF coil legs). All design
variants are required to provide a net electric output of 1000~MWe. The effect
of reduced efficiency is to increase the size of both in-vessel and ancillary
systems in order to produce the additional thermal power to reach
1000~MWe. Reduced power density mainly affects the size of the in-vessel
components and coils. The system optimization also maintained a fixed aspect
ratio $A=4$, plasma $\beta=3.4\%$ and peak allowable field at the TF coil of
16~T. Taking advantage of the low neutron flux in the shield, the original \AII
design was adjusted by replacing the 15\% V structure in the outer layers of
the shield by steel. A 5\% V coolant cladding is kept for coolant
compatibility.  By making this change, the cost of the vanadium in the shield
(which represents a major cost item) has been reduced substantially as compared
with earlier designs. The structures in the first wall, blanket, and reflector
are still made entirely of vanadium. The presence of steel in the shield is not
expected to prevent the plant from meeting the Class-C waste disposal or safety
requirements.

To simulate a steel design in the ASC, the V structure simply was replaced with
ferritic steel. The U.S.\ has not developed a serious power plant design using
ferritic steel in many years (since STARFIRE~\cite{STARFIRE}).  Recent work in
Europe~\cite{Dal95,Gia95} was used to determine the appropriate range of
thermal conversion efficiencies to use. It is true that there will be many
differences between the ferritic-steel and V designs in terms of radial
dimensions, energy multiplication, pumping power, power conversion system,
level of safety assurance (LSA) rating, {\it etc.}, but these differences
(which include both positive and negative effects) were not factored into this
analysis.

Different options for steel power plants were assessed with simple economic
models that have net thermal efficiencies of 30\% and 35\%.  A 35\% thermal
efficiency might be expected with a helium-cooled system, whereas a lower
efficiency of 30\% corresponds more closely to a water cooled (with Li$_2$O
breeder) system~\cite{Dal95,Gia95}.  Higher efficiencies may be possible with
advanced designs, but the analysis here is based on existing European Demo
blanket designs. 

The lower power density of steel has been enforced by limiting the peak neutron
wall loading to $\leq4$~MW/m$^2$. This limitation is design-dependent, and is
based on engineering judgment. Perhaps more importantly, first-wall surface
heat flux is constrained to $\leq0.5$~MW/m$^2$
(\hbox{Table~\ref{t-6-2-2}}). Based on the \AII reference design, peaking
factors on the wall loading and surface heat flux of 1.75 and 1.2 were used,
respectively. The surface heat flux peaking factor assumes peaking of only the
bremsstrahlung heat flux (edge plasma heat loads on the first wall and divertor
are still highly uncertain). For vanadium, engineering limits on power density
are not reached in the reference design.

The end-of-life first-wall fluence used for the comparison is 15~MW-y/m$^2$ for
both ferritic steels and V.  This is based on a 200~dpa limit. These values are
highly uncertain, given the absence of high-fluence data. In addition, they are
likely to be dependent on the operating conditions, such as stress
levels. Given a fixed fluence lifetime, the component operating life depends
only on the wall loading. Operating lifetime affects the replacement cost and
also the availability (less frequent replacement lowers the annual replacement
cost and increases the expected availability).  Operation at lower power
density results in slower neutron damage accumulation, thus extending the
operating life of components. Assuming the time to repair and replace
components is not changed, this leads to higher availability. The cost of
electricity (COE) is inversely proportional to the plant availability factor,
which is expected to be dominated by the planned outages for regular in-vessel
component replacement. Therefore, one expects a lower power density system to
benefit from this effect with a reduced COE. A second, but more subtle effect
on availability results from the fact that systems pushed to higher power
density are likely to suffer from higher failure rates. This effect has been
ignored in this analysis, as it is difficult at this time to quantify such an
effect.

The gross thermal conversion efficiency used here is the ratio of the gross
electric power divided by the total useful thermal power generated in the
fusion power core. It does not include plant recirculating power factors. The
key operating parameters which determine the maximum achievable gross
efficiency are the maximum allowed material design temperatures and stresses.
Lower efficiencies for ferritic steels result from both lower temperature
limits and lower thermal stress factors.

For the coolant options under consideration, the V/Li self-cooled design
utilizes an intermediate heat exchanger for the control of tritium leakage, and
is coupled to a Rankine cycle. Efficiencies as high as 49\% have been
predicted, but a more conservative value of 46\% is adopted here assuming a
lithium coolant outlet temperature of 610\SdegC. For the helium-cooled option,
a similar gross thermal efficiency can be reached with the Rankine cycle.  Even
though a secondary loop is not required, the maximum Rankine cycle efficiency
is controlled by the thermodynamic properties of the steam side.

For the ferritic steel designs, the thermal conversion efficiency was varied
parametrically to determine the minimum goals for COE to be comparable with V
systems. It is very important to consider the loss of useful thermal power in a
ferritic steel divertor. The divertor power accounts for up to 25\% of the heat
removed from the power core, including neutron, thermal radiation, and
transport power.  High-heat-flux components using steel are likely to be
constrained to operate with low coolant temperature, thus reducing the gross
thermal efficiency of the plant substantially.

\begin{table}[b]
\caption{}
\centering{\bf Major Parameters of Modified \AII Design} 
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l @{\hspace{1cm}}  c}
\hline \\
Net electric power (MW)                      & 1,000 \\
Useful thermal power (MW)                    & 2,570 \\
Gross thermal efficiency                     & 46\% \\
Major radius (m)                             & 5.60 \\
Peak/average wall load (MW/m$^2$)            & 5.1/2.9 \\
Peak/Average first-wall heat flux (MW/m$^2$) & 0.37/0.31 \\
On-axis toroidal field (T)                   & \Z8.0 \\
Maximum field at the coil (T)               &  15.9 \\
Fusion power density (MW/m$^3$)              & 4.8 \\
First wall and blanket cost (M\$)            & \Z\Z54 \\
Vacuum vessel cost (M\$)                     & \Z\Z51 \\
Shield cost (M\$)                            & \Z256 \\
Total direct capital cost (M\$)              & 2,048 \\
First wall and blanket replacement cost (mill/kWh) & 4.0 \\
Cost of electricity (mill/kWh)              & 71 \\
\\ \hline
\end{tabular}
\label{t-6-2-3}
\end{table}

Table~\ref{t-6-2-3} lists the major device parameters adopted from \AII. The
COE for the modified V system (with steel used in the shield) is 3~mill/kWh
less than the previous reference case (with vanadium structure throughout).
Table~\ref{t-6-2-4} summarizes the economic results for the ferritic steel
cases. In the case with unconstrained wall loading, the COE is competitive with
the V plant at a reasonable efficiency of 35\%. However, in order to account
for the lower conversion efficiency, the code has optimized by increasing the
peak power density to over 5~MW/m$^2$ (and peak surface heating to nearly
0.5~MW/m$^2$).

\begin{sidewaystable}[p]
\caption{}
\centering{\bf Impact of Gross Thermal Efficiency on the Performance of the 
Ferritic Steel Designs}
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l c c c c c c c c c }
\hline 
                   & \multicolumn{4}{c}{Unconstrained neutron wall loading} &&
 \multicolumn{4}{c}{Constrained neutron wall loading} \\
\cline{2-5} \cline{7-10}
Gross thermal efficiency $^{(b)}$    & 0.25   & 0.30   & 0.35   & 0.40   &&
 0.25   & 0.30   & 0.35   & 0.40   \\
\hline 
         & \hspace{1.2cm} & \hspace{1.2cm} & \hspace{1.2cm} & \hspace{1.2cm} &&
    \hspace{1.2cm} & \hspace{1.2cm} & \hspace{1.2cm} & \hspace{1.2cm} \\
Net electric power (MW)              & 1,000  & 1,000  & 1,000  & 1,000  &&
 1,000  & 1,000  & 1,000  & 1,000  \\
Major radius (m)                     & 6.52   & 6.24   & 6.00   & 5.80   &&
 9.08   & 8.16   & 7.48   & 6.96   \\
Peak wall load (MW/m$^2$)            & 7.7    & 6.9    & 6.2    & 5.7    &&
 4.0    & 4.0    & 4.0    & 4.0    \\
Peak first-wall heat flux (MW/m$^2$) & 0.57   & 0.50   & 0.46   & 0.42   &&
 0.29   & 0.30   & 0.30   & 0.30   \\
On-axis toroidal field (T)           & \Z8.5  & \Z8.4  & \Z8.2  & \Z8.1  &&
 \Z6.6  & \Z6.8  & \Z7.0  & \Z7.1  \\
Maximum field at the coil (T)        &  15.9  &  15.9  &  15.9  &  16.0  &&
  11.2  &  11.9  &  12.5  &  13.0  \\
Fusion power density (MW/m$^3$)      & 6.2    & 5.8    & 5.4    & 5.2    &&
 2.3    & 2.6    & 2.8    & 3.0    \\
First wall and blanket cost (M\$)    & \Z\Z26 & \Z\Z24 & \Z\Z22 & \Z\Z21 &&
 \Z\Z43 & \Z\Z37 & \Z\Z32 & \Z\Z28 \\
Vacuum vessel cost (M\$)             & \Z\Z69 & \Z\Z64 & \Z\Z59 & \Z\Z55 &&
 \Z132  & \Z107  & \Z\Z90 & \Z\Z78 \\
Shield cost (M\$)                    & \Z261  & \Z242  & \Z226  & \Z214  &&
 \Z443  & \Z377  & \Z326  & \Z289  \\
Total direct capital cost (M\$)      & 2,398  & 2,206  & 2,067  & 1,975  &&
 2,718  & 2,431  & 2,235  & 2,098  \\
Replacement cost$^{(b)}$ (mill/kWh)
                                     & 3.8    & 3.0    & 2.5    & 2.1    &&
 2.9    & 2.4    & 2.1    & 1.9    \\
Cost of electricity (mill/kWh)      & 81     & 74     & 70     & 67     &&
 89     & 80     & 74     & 70     \\
\\ \hline
\multicolumn{10}{l}{$(a)$ Including the pumping power.} \\
\multicolumn{10}{l}{$(b)$ For first wall, blanket, and reflector.}
\end{tabular}
\label{t-6-2-4}
\end{sidewaystable}

\clearpage

The effect of imposing a wall-loading constraint on the ferritic-steel designs
is to increase the size by 25\%--30\% and the COE by 6.2\%--7.7\%, as shown in
\hbox{Table~\ref{t-6-2-4}}. Comparing all designs at a common LSA~=~2 rating, a
wall-loading constrained steel design must attain a 39\% thermal conversion
efficiency to have a COE comparable to that for the modified V design.  A 34\%
gross thermal efficiency (ratio of gross electric to total thermal power) is
required for the unconstrained wall-loading steel design to be comparable in
cost to the improved V design. If the ferritic steel system is unable to meet
the LSA~=~2 rating, the difference between the COE of the FS and V systems will
become larger.  The full range of cases examined here have a net COE between
70--80~mill/kWh. This is a significant, but not definitive difference. In the
modified \AII design, the fusion power core cost does not dominate the COE;
therefore savings obtained by further minor adjustments are not expected to
save more than a few mill/kWh.

\subsubsection{Activation of ferritic steels}
\label{s-6-2-4-4}

The safety characteristics of fusion power plants are directly related to the
choice of structural materials.  Low activation ferritic/martensitic steels are
the most attractive alloys because they produce only low-level waste (Class C)
at the end of the life.  Detailed activation analyses were performed to
identify the safety, environmental and rad-waste characteristics of five
different ferritic steel alloys: \HT, low activation (modified) \HT, the IEA
modified F82H steel (\hbox{9Cr-2WVTa}), and the leading European Community
candidates, MANET-2, and LA12TaLC alloys. Several activation-related issues
were investigated for each of the steel alloys: activity, decay heat, and
biological hazard potential (BHP). The waste disposal ratings (WDR) of each of
the alloys at the end of its lifetime were also evaluated.

\paragraph{Structure Activity, Decay Heat, and Biological Hazard Potential.}

The neutron flux used for the activation calculation was generated by the
one-dimensional discrete ordinates neutron transport code
ONEDANT~\cite{ONEDANT}.  A 46-group neutron and 21-group gamma coupled
cross-section library containing P$_3$ Legendre expansions of the scattering
cross sections based on the ENDF/B-V basic data files was used in the transport
calculations.  The activation calculations were conducted using the DKR-ICF
computer code~\cite{DKR-ICF} with activation cross sections taken from the new
USACT93~\cite{REAC3}.  The USACT93 library is based on neutron transmutation
cross section and isotopic radioactive decay data from the ENDF/B-VI and EAF3
files.  The neutron transmutation data is in a 46-group structure format.  All
materials were assumed to stay in the reactor for 3 full power years (FPY).
The biological hazard potentials were calculated using the maximum permissible
concentration limits in air for the different isotopes according to the NRC
regulations specified in 10CFR20~\cite{10CFR20}. A scalable average neutron
wall loading of 1~MW/m$^2$ was used in the analysis.

Figures~\ref{f-6-2-03} to \ref{f-6-2-08} shows that all of the ferritic and
martensitic steel alloys produce identical amount of radioactivity during the
first 10 years following shutdown.  The short-term radioactivity induced in
steel is generally dominated by $^{56}$Mn ($T_{1/2}=2.6$~h), $^{54}$Mn
($T_{1/2}=312$~d), $^{51}$Cr ($T_{1/2}=27.7$~d), and $^{55}$Fe
($T_{1/2}=2.73$~y).  The intermediate-term radioactivity is dominated by
$^{55}$Fe, $^{54}$Mn, $^{60}$Co ($T_{1/2}=5.27$~y), and $^{3}$H
($T_{1/2}=12.7$~y).  The $^{55}$Fe and $^{54}$Mn isotopes, which are produced
by the transmutation of iron, dominate the radioactivity during the first few
years following shutdown. The long-term radioactivity is dominated by $^{53}$Mn
(\hbox{$T_{1/2}=3.7\X 10^6$~y}), $^{14}$C ($T_{1/2}=5,730$~y), $^{93}$Nb
($T_{1/2}=16.1$~y), and $^{94}$Nb (\hbox{$T_{1/2}=2\X 10^4$~y}).  The two
ferritic alloys, IEA modified F82H and modified \HT, as well as the low
activation martensitic alloy, LA12TaLC, result in lower levels of long-term
radioactivity due to their low niobium impurity content.

\begin{figure}[p]
\centerline{\psfig{file=sti-6-2-03.ps,height=3.25truein}}
\caption[f-6-2-03.ps]
{Activities induced in different ferritic steel candidates.}
\label{f-6-2-03}
\vspace{1cm}
\centerline{\psfig{file=sti-6-2-04.ps,height=3.25truein}}
\caption[f-6-2-04]
{Activities induced in the F82H alloy and the leading martensitic steel
candidates.}
\label{f-6-2-04}
\end{figure}

\begin{figure}[p]
\centerline{\psfig{file=sti-6-2-05.ps,height=3.25truein}}
\caption[f-6-2-05]
{BHP induced in different ferritic steel candidates.}
\label{f-6-2-05}
\vspace{1cm}
\centerline{\psfig{file=sti-6-2-06.ps,height=3.25truein}}
\caption[f-6-2-06]
{BHP induced in the F82H alloy and the leading martensitic steel candidates.}
\label{f-6-2-06}
\end{figure}

\begin{figure}[p]
\centerline{\psfig{file=sti-6-2-07.ps,height=3.25truein}}
\caption[f-6-2-07]
{Decay heat induced in different ferritic steel candidates.}
\label{f-6-2-07}
\vspace{1cm}
\centerline{\psfig{file=sti-6-2-08.ps,height=3.25truein}}
\caption[f-6-2-08]
{Decay heat induced in the F82H alloy and the leading martensitic 
steel candidates.}
\label{f-6-2-08}
\end{figure}

\paragraph{Rad-waste Classification.}

The structure activation results were utilized in the rad-waste classification
performed using the WDR computer code~\cite{WDR}.  The rad-waste of all alloys
was evaluated according to both the NRC 10CFR61~\cite{10CFR61} and
Fetter~\cite{Fet90} waste disposal concentration limits.  

There are three different classes of rad-waste under U.S.\ regulations.  The
most stringent requirements are for Class~A, segregated waste.  The major
hazard of this class of waste is to the individuals responsible for handling
it.  However, such waste is not considered to be hazardous following the loss
of institutional control over the disposal site.  The 10CFR61 regulations
assume that the waste-disposal site will remain under administrative control
for 100~years and that the dose at the site due to an inadvertent intruder at
the end of the 100~year period is less than 500~mrem/y.

Class~C intruder waste allows for higher concentrations of rad-waste.  Class~C
waste must be packaged and buried such that it will not pose a hazard to an
inadvertent intruder after the 100~year institutional period is over.  Class~C
waste is assumed to be stable for 500~years.

The waste-disposal rating (WDR) is defined as the sum of the ratio of the
concentration of a particular isotope to the maximum allowed concentration of
that isotope taken over all isotopes and for the particular class.  If the
calculated WDR $\leq1$ when Class~A limits are used, the rad-waste should
qualify for Class~A.  If the WDR is $>1$ when Class~A limits are used but
$\leq1$ when Class~C limits are used, then the waste is termed Class~C .  If
WDR $>1$ when Class-C limits are used, it implies that the rad-waste does not
qualify for shallow land burial.

Fetter developed a modified version of the NRC intruder model to calculate
waste disposal limits for a wider range of long-lived radionuclides that are of
interest for fusion researchers than the few that currently exist in the
current 10CFR61 regulations.  Fetter's model included more recent transfer
coefficients and dose conversion factors.  However, while the NRC model limits
the whole body dose to 500~mrem or the dose to any single organ (one of seven
body organs) to 1.5~rem, Fetter limits are based on the maximum dose to the
whole body only.

\begin{table}[b]
\caption{}
\centering{\bf Waste Disposal Ratings for Several Ferritic/Martensitic
Steels$^{\mathbf (a,b)}$}
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l c c c c c}
\hline
                &  HT-9             &  MHT-9             &  F82H              &
          MANET-2         &  LA12TaLC \\
\hline \\
Class A$^{(c)}$ & $1.83\X 10^3$     & 9.56               & 9.93               &
        $2.43\X 10^3$     &  90.3  \\
                & $^{94}$Nb (97\%)  & $^{3}$H (60\%)     & $^{3}$H (50\%)     &
        $^{94}$Nb (99\%)  & $^{94}$Nb (87\%)  \\  
Class C$^{(c)}$ & 178               & 0.187              &  0.156             &
          240             &  8.12  \\
                & $^{94}$Nb (99\%)  & $^{94}$Nb (92\%)   & $^{94}$Nb (55\%)   &
        $^{94}$Nb (100\%) & $^{94}$Nb (97\%)  \\
Class C$^{(d)}$ & 196               & 6.62               & 9.2                &
              264         &  7.92  \\
                & $^{94}$Nb (90\%)  & $^{192m}$Ir (70\%) & $^{108m}$Ag (98\%) &
        $^{94}$Nb (99\%)  & $^{94}$Nb (99\%)  \\
\\ \hline
\multicolumn{6}{l}{$(a)$ For first wall exposed to 1~MW/m$^2$ neutron wall load
for 3~FPY.}\\
\multicolumn{6}{l}{$(b)$ In each case, the dominant isotope and its
            contribution to WDR is given.}\\
\multicolumn{6}{l}{$(c)$ Based on 10CFR61 regulations.}\\
\multicolumn{6}{l}{$(d)$ Based on Fetter's estimates.}
\end{tabular}
\label{t-6-2-5}
\end{table}

Specific activities calculated by the DKR-ICF code were used to calculate
the waste disposal ratings for each alloy.  The waste disposal ratings for
Class~A and Class~C low-level waste are shown in \hbox{Table~\ref{t-6-2-5}}.
The 10CFR61 Class~A WDR is given after a waiting period of about 10~years to
allow for the specific activity of short-lived nuclides ($T_{1/2}\leq5$~y) to
drop below 7,000~Ci/m$^3$.  The 7,000~Ci/m$^3$ limit is 10 times larger than
the limit specified by the NRC for Class~A disposal of short-lived nuclides of
an unspecified waste form.  In comparison with other isotopes for which limits
are given for different waste forms, the factor of 10 is used for isotopes
contained in metal waste.  Since the NRC regulations do not specify any limit
for short-lived activity for Class~C waste, the Class~C WDR values were
calculated after a 1-year cooling period for both 10CFR61 and Fetter limits.

Table~\ref{t-6-2-5} shows that the IEA modified F82H and the modified \HT
alloys qualify for Class~C according to the NRC limits.  None of the steel
alloys would qualify for disposal as Class~C according to Fetter limits.
However, reducing the Ir and Nb impurities in MHT-9 and Ag in F82H by a factor
of $>10$ would qualify both steels as class C waste according to both NRC and
Fetter's limits.  Finally, it is important to keep in mind that the waste
disposal ratings given in \hbox{Table~\ref{t-6-2-5}} are only for the first
wall.  The corresponding blankets should have lower waste disposal ratings, as
they are exposed to lower neutron flux.  Disposing the first wall and blanket
together should also allow for lower waste disposal ratings than those given in
the table.

\subsection{Key Issues and R\&D Needs}
\label{s-6-2-5}

Two issues were selected for special attention due to their high degree of
uncertainty and/or large impact on the system attributes: loss of ductility and
its consequences, and ferromagnetism issues.  As a result of the analysis,
ductility was viewed as the most serious concern.  Ferromagnetism causes
important operational problems, but these were viewed as design dependent and
soluble.

\subsubsection{Loss of ductility and its consequences}
\label{s-6-2-5-1}

There are three types of ductility that are used in assessing material
performance.  The first type that is commonly used in design codes is the
tensile ductility, which is measured in the standard tensile test at modest
strain rates.  A particular measure of ductility, uniform elongation (UE), is
used in design codes to classify materials according to ductile (\hbox{${\rm
UE} \geq 5\%$}), semi-brittle (\hbox{$1 < {\rm UE} < 5\%$}) or brittle
(\hbox{${\rm UE} \leq 1\%$}).  A second type of ductility is the measure of the
uniform strain at failure in a standard creep test, which generally involves
strain rates much lower than those used in tension tests.  This parameter is
important for high temperature operation.  The third type of ductility is a
measure of the ability of a material to absorb impact loads.  As such, it
involves high strain rates.  This third type really should be expressed as a
strain energy or fracture toughness.  However, as the Charpy test is often
used, the information obtained is usually given as a ductile-to-brittle
transition temperature (DBTT) and an upper-shelf energy ({\it i.e.}, the
maximum energy absorbed when the material is behaving in a ductile mode).

Correlations have been established between Charpy test results and fracture
toughness for austenitic stainless steels, but not for ferritic steels.  In
design codes, the fracture toughness is used to limit the peak stress in the
regions of stress concentration.  The Charpy test is used to define a
nil-ductility temperature which is similar to, but not equal to, the DBTT.  The
ASME code simplistic rules require that steel pressure vessels be operated at a
temperature (dependent on assumed flaw size) greater than the nil-ductility
temperature.  The Charpy test is also used to monitor the change in ductility
of surveillance samples of pressure vessel material as the irradiation
progresses.
In the following, the ferritic-steel ductility data obtained from tensile tests
(UE) and Charpy tests (DBTT) are reviewed.  The implications of such data are
also discussed.

\paragraph{Tensile-Test Ductility.}

The uniform elongation of unirradiated \hbox{Fe-9Cr-1MoVNb} and
\hbox{Fe-12Cr-1MoVW} generally decreases with increasing temperature from
5\%--10\% at room temperature to 1\%--3\% at 550\SdegC
(see \hbox{Fig.~\ref{f-6-2-09}}).  Above 550\SdegC, the uniform elongation
increases as the material softens.  These values are low compared to austenitic
stainless steels (20\%--30\%) in the same temperature range.  However, in
reviewing the details of the ASME code design rules, such low ductility would
not result in a reduction of the allowable design stresses.

Exposure to a neutron environment, particularly in the temperature range of
25--400\SdegC, results in a loss in ductility of these alloys
(\hbox{Fig.~\ref{f-6-2-10}}).  The decrease in ductility appears to be more
significant when the materials are tested in mixed-spectrum reactors,
particularly if extra Ni has been added to enhance He production.  The data in
\hbox{Fig.~\ref{f-6-2-10}} indicated by $<1$~He/dpa (1~appm He/dpa) are
fast-reactor data, while all other points are obtained after irradiation in
mixed-spectrum reactors.  The decreases in ductility below 1\% would require
very careful design analysis to determine if a reduction in allowable stresses
would be warranted.

\begin{figure}[p]
\centerline{\psfig{file=sti-6-2-09.ps,height=3.truein}}
\caption[f-6-2-09]
{Uniform elongation as a function of temperature for unirradiated ferritic
steel alloys \hbox{Fe-12Cr-1MoVW} and \hbox{Fe-9Cr-1MoVNb}.  Results include
many different heat treatments and variations in the details of the chemical
composition.}
\vspace{1.0cm}
\label{f-6-2-09}
\centerline{\psfig{file=sti-6-2-10.ps,height=3.truein}}
\caption[f-6-2-10]
{Uniform elongation as a function of temperature for irradiated ferritic steel
alloys \hbox{Fe-12Cr-1MoVW} and \hbox{Fe-9Cr-1MoVNb}.  Results include many
different heat treatments and variations in the details of the chemical
composition.}
\label{f-6-2-10}
\end{figure}

With regard to the uniform elongation of the reduced-activation
\hbox{Fe-9Cr-2WVTa} alloys, most of the data reported in the literature pertain
to total, rather than uniform, elongation which makes comparisons difficult.
However, one reported data set shows that after irradiation in a fast-neutron
spectrum (FFTF) at 420\SdegC, the uniform elongation is 3\%--8\%.  These tests
would not include the effects of any significant He generation, nor would they
be indicative of the behavior at lower temperatures.

\paragraph{Ductile-to-Brittle Transition Temperature (DBTT).}

The DBTT for unirradiated \hbox{Fe-12Cr-1MoVW} alloys falls into the range of
--50 to 0\SdegC.  For \hbox{Fe-9Cr-1MoVNb}, the range is --80 to 20\SdegC.  The
results are highly dependent on the thermo-mechanical heat treatment used
during fabrication and the size and shape of the specimen.  The
\hbox{Fe-(7--9)Cr-WVTa} have some of the lowest DBTTs reported for ferritic
steels: --40 to --90\SdegC, which is part of the reason for such optimism
regarding these materials.

\begin{figure}[b]
\centerline{\psfig{file=sti-6-2-11.ps,height=3.25truein}}
\caption[f-6-2-11]
{Ductile to Brittle Transition Temperatures (DBTT) for Fe-12Cr-1MoVW,
Fe-9Cr-1MoVNb and reduced-activation Fe-9Cr-2WVTa ferritic steels irradiated to
neutron damage levels of $<43$~dpa and He contents $<400$~appm.  He contents
are in appm.}
\label{f-6-2-11}
\end{figure}

The DBTT values for irradiated materials are somewhat perplexing in that there
appears to be a strong correlation between the increase in DBTT and He
generation, as well as neutron damage measured in displacements per atom (dpa).
\hbox{Fig.~\ref{f-6-2-11}} shows a wide range of data with the
\hbox{Fe-12Cr-1MoVW} irradiated at 400\SdegC with 1--10~appm~He/dpa having a
DBTT value as high as 330\SdegC.  In general, most data points are below
250\SdegC.  It is encouraging to note that the reduced-activation
\hbox{Fe-9Cr-2WVTa} shows no significant increase in DBTT for irradiation up to
13 dpa with negligible He generation.  To add to the complexity of trying to
interpret the effects of helium generation, more recent data at $\leq$ 2.5 dpa
and $\leq3.5$~appm~He, show increases of 150--180\SdegC in DBTT after
irradiation at 300-400\SdegC for \hbox{Fe-12Cr-1MoVW} and 60--75\SdegC for
\hbox{Fe-9Cr-1MoVNb} under the same conditions~\cite{Klu95}. The resulting
absolute values of DBTT are 100--150\SdegC for the \hbox{Fe-12Cr-1MoVW} alloy
and 30--45\SdegC for the \hbox{Fe-9Cr-1MoVNb} alloy.

The design consequences of a high DBTT are serious because based on ASME
guidelines, the minimum operating temperature would have to be set at a higher
temperature, which is dependent on wall thickness and assumed undetected crack
length.  The issue of allowable stresses would have to be examined carefully to
determine the effects of reduced tensile ductility on operating load limits.
There would still be some concern about the behavior of such a material during
shut-down to room temperature.  In addition to shut-down stresses ({\it e.g.,}
reversals in thermal stresses and residual stresses caused by creep and plastic
deformation), there is the possibility of seismic or other loads during
shut-down.  Thus, although it is not a strict design code requirement, the
``safety margin'' increases significantly as the DBTT is decreased to below
room temperature.

There are two approaches to the DBTT problem that are being pursued.  The E.U.\
team is considering the possibility and desirability of heat treating the steel
periodically to 400--500\SdegC during shut-down to partially anneal out some of
the irradiation-induced hardening.  The other approach, which is the one being
adopted by the IEA Working Group, is to optimize the chemical composition and
thermo-mechanical heat treatment to decrease the irradiated DBTT value.
Success has been achieved in doing this up to 30~dpa at 365\SdegC in a
fast-reactor environment.  It remains to be demonstrated that the DBTT values
remain low to higher neutron damage levels with He generation and down to lower
operating temperatures.  Progress also can be realized in this area by using
fracture toughness data in place of Charpy DBTT data and the sophisticated ASME
design criteria in place of the simplistic rules.

\subsubsection{Ferromagnetism issues}
\label{s-6-2-5-2}

Another family of concerns regarding the use of ferritic steels in magnetic
fusion is related to the interaction of ferromagnetic materials with magnetic
fields.  The concerns include effects on field ripple, perturbations of the
poloidal field, and magnetic loads on the structure.  Each of these issues has
been investigated previously~\cite{Lec93} and the results of those previous
studies are summarized here.  In general these studies have concluded that
proper design can mitigate these issues, such that they should not be
considered as feasibility concerns, but rather R\&D issues.  These previous
studies did not consider certain important issues, so areas of future work are
identified at the end of the section to guide design studies which choose to
adopt ferritic steels as a structural material.

\paragraph{Field ripple.}

Field ripple (non-uniformity in toroidal field) occurs in all toroidal systems
because the toroidal field coils are discrete.  The concern with the use of
ferritic steel structures in tokamaks is that the ferromagnetic materials will
enhance the ripple.  If this increase is significant, particle losses will be
enhanced and the disruption frequency may be increased.

Lechtenberg \ea~\cite{Lec93} have conducted sample ripple calculations for
three conceptual designs: STARFIRE~\cite{STARFIRE}, TNS~\cite{TNS}, and
ETF~\cite{Bec80}.  These calculations incorporated a ferritic steel which had a
saturation magnetization of 1~T which was reached when the ambient field was
2~T.  A square blanket penetration of 0.5~m was included in the analysis.  For
these designs, the ripple peaked at around 0.2\% with a non-ferromagnetic
blanket and substituting a ferromagnetic blanket increased the peak to around
0.3\% (with the penetration included).  An axisymmetric ferromagnetic blanket
caused no discernible difference in the ripple.  It is the penetrations --
asymmetries -- that cause perturbations in the field ripple.  It was concluded
this level of increase in the ripple is not significant.

\paragraph{Poloidal field perturbation.}

A related effect of the use of ferromagnetic materials in tokamaks is the 
perturbation of the poloidal field.  Lechtenberg \ea~\cite{Lec93} found that 
poloidal field perturbations peaked at approximately 3\% when a ferromagnetic 
blanket was added to STARFIRE.  They concluded that this is excessive but 
active control of the poloidal field coils could alleviate this problem.  They 
further concluded that this would not add complexity to the tokamak design 
because tokamaks would probably include active poloidal field control for 
plasma stability and position control.  An additional concern here is the 
delay of the control field penetration caused by the ferromagnetic structure.  
As before, this concern is thought to be alleviated by the saturation of the 
steel magnetization.  This issue is included in the list of action items at 
the end of this section, because there is no clear evidence that a 3\% 
perturbation in the poloidal field is acceptable.  Realistic limits for 
poloidal field variations must be determined.

\paragraph{Ferromagnetic loads.}

The use of ferromagnetic materials in tokamaks introduces a set of loads that
are not an issue for other materials.  These are the magnetic loads induced on
the structure by the magnetic fields in the device.  Simple, cylindrical models
of STARFIRE and ETF have been derived; they predict static magnetic pressures
of approximately 0.5~MPa in the ferromagnetic structure.  This should be easily
manageable in a tokamak blanket.  Unfortunately, the issue cannot be dismissed
this easily.  Of greater concern are the forces induced on pipes and ducts
which pass through regions of high field gradients.

Lechtenberg \ea~\cite{Lec93} calculated the stresses on a coolant pipe passing
between the outer legs of the toroidal field coils.  The magnetization field
was calculated using the GFUN3D code and the stresses were calculated using
MODSAP, a finite element code.  The pipe was assumed to be 4 m long, with an
8.25~cm outer diameter and a 0.75~cm wall thickness.  The material used was
\HT.  The distance from the pipe to the center of the adjacent magnet was
varied from 80 to 140~cm.  At a distance of 86~cm from the magnet center, the
calculated force in the pipe peaked at approximately 7,000~N/m, with the force
acting transverse to the pipe and with the peak force occurring about 50~cm
from one end of the pipe.

The stresses induced in the pipe by these forces are calculated to be
approximately 175~MPa.  This is a substantial stress for \HT structures.
Fortunately this stress can be reduced by providing additional supports to the
pipe.  In fact, addition of a single support at an optimal location in the pipe
reduces this stress to less than 30~MPa.  This is a reasonable stress and
should not appreciably affect the life of this component.  These forces and
their associated stresses can be further reduced by moving the pipe further
from the center of the adjoining magnet.

\paragraph{Issues Requiring Further Study}

If future design studies consider the use of ferritic steels, the following 
issues should be clarified in order to ensure reliable and effective use of 
these materials:

\begin{itemize}

\item
The effect of realistic blankets on field ripple should be assessed.  This 
should include asymmetries and realistic penetrations.

\item
The allowable perturbation in the poloidal field should be determined.

\item The interaction of the transient magnetic fields, {\it e.g.,} during
start-up and disruptions, with the ferromagnetic structure should be assessed
({\it e.g.,} forces, delays in field penetration, impact on the vertical
stability feed-back system).

\item
The loads and stresses on a realistic blanket and associated ducting, piping, 
and manifolds should be re-evaluated.  This can be done with existing, 
commercial finite element codes.

\item
The stored energy associated with the magnetization of these materials and 
their effects on source terms for safety calculations should be assessed.

\end{itemize}


\section{VANADIUM DESIGNS}
\label{s-6-3}

\subsection{Overview}
\label{s-6-3-1}

Vanadium alloys have been included in many fusion R\&D programs throughout 
the world due to their potential for improved thermomechanical performance, 
safety advantages, and possibly longer lifetime.  As with ferritic steel 
designs, our assessment considered the ability of vanadium-based designs to 
meet the essential mission requirements of a fusion Demo, as well as the 
uncertainties which currently exist.

The use of vanadium restricts the use of some materials (such as for the
coolant and breeder) due to compatibility.  For example, water is not
considered a viable candidate as the coolant.  Design options using both
lithium and helium as coolant are described and used for the assessments of the
first-wall and blanket, divertor, shielding and heat transport loop.

The assessment of blanket options concentrated on the use of lithium as
breeder.  Two coolant options were considered: self-cooled lithium and
high-pressure helium.  Several previous design
studies~\cite{BCSS,ARIES4,UWMAK1,UWMAK3,Ehs87,TITAN} have provided detailed
descriptions of the self-cooled concept, such that only a cursory review of the
design concept is repeated here.  The He-cooled blanket option is relatively
less studied, and therefore a much more thorough description of that class of
design concepts has been provided.  Liquid Pb-Li is a possible candidate for
the breeder and/or coolant, but U.S.\ R\&D programs do not currently support
this option.

Three power plant requirements were selected for more detailed examination.
First, the estimated high cost of vanadium is a major concern.  The improved
performance must offset the higher capital and replacement costs for the
in-vessel components (as compared to steels).  Strategies for minimizing the
quantity of vanadium, maximizing the lifetime and replacing (or recycling)
components are addressed.  Second, accident safety and waste management are
examined to determine whether the Demo requirements can be met.  Finally,
reliability and availability are briefly surveyed.  All design concepts will
have a difficult development path to meet the very stringent utility
requirements on availability.

Three of the critical issues associated with the use of vanadium have been
examined.  The materials data base for vanadium alloys is small, but growing.
An extensive review of the state of our knowledge of vanadium properties is
provided in Sec.~\ref{s-6-3-3} and in the Appendix.  Some of the important
material-related issues include tensile and creep behavior under irradiation,
compatibility with coolants and breeders, and fabrication.  Tritium recovery
and control is the second issue examined.  Recovery methods from lithium have
been developed, but the stringent requirements for impurity control (to prevent
degradation of the vanadium properties) leads to special concerns.  Finally,
the possibility of vanadium recycling was explored as a possible means to
offset the high cost of the material.

While uncertainties in the potential attractiveness of vanadium systems is
acknowledged, it was decided to develop the V/Li system for the detailed
conceptual design phase which followed these assessments.  The ultimate
potential of fusion depends on our ability to provide high performance systems
with minimal safety and environmental impacts.  Vanadium systems appears to
offer the chance to meet this challenge.

\subsection{Design Options}
\label{s-6-3-2}

Several systems in the fusion power core may use vanadium structure, including
the blanket and first wall (Sec.~\ref{s-6-3-2-1} through \ref{s-6-3-2-3}),
divertor and other plasma facing components (Sec.~\ref{s-6-3-2-4}), shield
(Sec.~\ref{s-6-3-2-5}), and primary heat transport loop (Sec.~\ref{s-6-3-2-6}).

\subsubsection{Self-cooled lithium blanket design}
\label{s-6-3-2-1}

The self-cooled lithium breeder blanket with a vanadium alloy structure has
been evaluated as the reference design of many
studies~\cite{BCSS,ARIES4,UWMAK1,UWMAK3,Ehs87,TITAN}.  This concept provides
several inherent characteristics that enhance the attractiveness of fusion as
an energy source.  Since lithium serves as both breeder and coolant, a simple
integrated first wall/blanket design configuration with a single pass,
poloidally oriented coolant flow has been developed.  This concept makes use of
an electrically insulated coolant channel to reduce the MHD-related pressure
drops.  This provides for a low pressure blanket and heat transport system
which, combined with the simple configuration, should substantially contribute
to high reliability.  All in-core complexity related to tritium recovery is
eliminated since tritium can be recovered from a small side stream off the
external coolant loop.  Also, concerns relating to coolant-breeder interaction
are not an issue with lithium both as breeder and coolant.

The lithium-vanadium system also provides several inherent safety features.
Because of the very high solubility of tritium in lithium, containment of
tritium during normal operation does not appears to be a critical issue.  Also,
the relatively high tritium concentration in lithium ($\sim1$~appm) results in
a modest size of tritium recovery system.  The low blanket pressure ($<1$~MPa
for design with insulator coating), high heat capacity, and thermal
conductivity of lithium and relatively low afterheat in this system provides
substantial advantages in the event of a loss-of-flow accident (LOFA).  Design
solutions developed to improve safety include the use of an inert gas to
eliminate lithium-air reaction and the use of a lithium-cooled shield and
in-vessel components to eliminate potential lithium-water reaction.

As in all systems, the vanadium-lithium concept exhibits a few inherent
characteristics that must be accommodated by special design solutions to
provide an attractive engineering system. Dominant among those are:
(1)~Chemical reactivity of lithium with air and water, and (2)~Liquid-metal MHD
effects. Lithium at high temperatures will react with water and air.  This is
the inherent chemistry of lithium and cannot be changed.  However, there are
design solutions to minimize the occurrence of those accidents.  The use of
inert cover gas and the elimination of water from the power plant core will
reduce the possibility that these reactions will occur. Even if those reactions
happened, the consequence may be acceptable.  Detailed analyses of lithium
fires have been carried out in the past using the LITFIRE code.  The conclusion
was that the temperature would be raised to $\sim$1,000\SdegC in a lithium-air
fire.  All safety evaluations were based on this temperature excursion.  Even
with those conditions, a high safety rating was achieved.

The MHD effects associated with the motion of a conducting liquid metal in a
conducting coolant channel result in a pressure drop.  This pressure drop is
proportional to the wall thickness.  Thus, increasing the structural fraction
to accommodate the pressure will be of no use, since the pressure drop only
increases further. In typical tokamak power plant designs, the MHD pressure
drop for the inboard regime of a tokamak results in pressure stresses beyond
the material limits.  To reduce this MHD pressure drop, some form of insulating
wall is required, either with a direct coating or with a sandwich-type
structure.  The reference design for the U.S.\ ITER breeding blanket and \AII
both have a direct coating.  This coating development is still in a very early
stage.

The cross-section of a Li/V blanket is shown in \hbox{Fig.~\ref{f-6-3-01}}.
The first wall is a continuous 2-mm thick plate.  The first wall coolant
channels are rectangular ducts 10~mm $\times$ 50~mm in cross section.  The
radial direction of the first wall channel must be made as small as possible to
improve heat transfer.  (If MHD turbulence can  improve heat transfer, the
dimension of the first wall channel could be relaxed.)  The radial dimension of
the back channels can be much larger.  The velocity in the breeding zone ducts
decreases with increasing radial position.  This provides a
uniform coolant exit temperature.  Tailoring of the velocity to the
location of the zone can be accomplished with ducts whose radial dimension
increases with distance from the first wall.

\begin{figure}[t]
%\vspace*{4.2in}
\centerline{\psfig{file=sti-6-3-01.ps,height=3.5truein,angle=-90}}
\caption[f-6-3-01]
{Cross-section of a self-cooled, lithium/vanadium blanket.}
\label{f-6-3-01}
\end{figure}

A major advantage of a self-cooled system, as depicted in
\hbox{Fig.~\ref{f-6-3-01}}, is that there is no heat transfer surface inside
the blanket.  The fusion energy is directly deposited within the coolant, that
the design of a heat exchanger within the confines of the blanket geometry is
not required.  Also, the only pressure containment surface will be the first
wall.  Thus, the frequency of coolant leakage can be greatly reduced.  This
effect will be discussed later in this report.

\subsubsection{Helium-cooled blanket design}
\label{s-6-3-2-2}

The motivation for the use of helium as the blanket coolant is its potential
for high thermal efficiency, safety, and waste disposal advantages.  Helium has
thermodynamic and transport properties that allow high temperature and hence
high efficiency operation.  It is a natural match with the Brayton power
conversion cycle.  Helium chemical inertness and non-activation significantly
reduce the risk of chemical safety and coolant waste disposal.  In order to
realize these advantages, the design must overcome the relatively low
volumetric specific heat of helium, which usually leads to a high-pressure
system.  When combined with the use of V-alloy, the design must also
successfully address the issue of coolant impurity control.

Three representative helium-cooled blanket designs have been reviewed for use
with V-alloy.  They are the helium-cooled ITER blanket design, the reference
Safety and Environmental Assessment of Fusion Power~\cite{Sal95} blanket
design and the helium-cooled, lithium breeder V-alloy (He-Li-V) design
developed for this study.

The helium-cooled, V-alloy blanket proposed for ITER~\cite{Won94} is similar to
that illustrated in \hbox{Fig.~\ref{f-6-3-02}}.  It is a poloidal flow co-axial
tube, molten Li-breeder design, with narrow modules that could fit the narrow
vertical maintenance opening of the 1993 ITER-EDA design.  This design focused
on addressing the question of reliability when high pressure helium (18~MPa) is
used.  All piping joints are located outside of the vacuum vessel.  Co-axial
tubes were used to increase design reliability.  Both tubes are designed to
take the full helium pressure and only the inner tube was pressurized.  The
annulus between tubes was filled with liquid metal (Na or Li) for leak
detection.  It addresses many of the design issues, but not the issue of
material compatibility between V-alloy and helium impurities.

\begin{figure}[b]
%\vspace*{4.2in}
\centerline{\psfig{file=sti-6-3-02.ps,height=4.0truein}}
\caption[f-6-3-02]
{The He-Cooled ITER Blanket Module which uses \HT as structural material (a
similar V-alloy design has been developed).}
\label{f-6-3-02}
\end{figure}

The second design is the reference design for the SEAFP reactor
study~\cite{Sal95}.  This is a poloidal flow, co-axial tube, Li$_2$O
breeder-in-tube (BIT) design, as shown in \hbox{Fig.~\ref{f-6-3-03}}.  As
indicated, the solid breeder Li$_2$O is also used to fill in the space between
the tubes.  Helium coolant flows in the annulus between the co-axial tubes.
The key focus of the SEAFP design is to ensure public acceptance and take into
account the operating requirements of utilities.  The issue of material
compatibility between Li$_2$O/V-alloy and helium impurities/V-alloy is proposed
to be resolved by a coating of sub-micron thick SiC or Al$_2$O$_3$.

\begin{figure}[b]
\vspace*{5.0in}
%\hbox to 6.25truein{\hfil
%\psfig{file=sti-6-3-03.ps,height=4.0truein}
%\hfil}
\caption[f-6-3-03]
{Reference SEAFP design.}
\label{f-6-3-03}
\end{figure}

The third design is cooled by helium and uses liquid lithium as the breeder
(\hbox{He-Li-V}).  This design is discussed in detail in the next section.
Selected parameters of \hbox{Ref-SEAFP} and \hbox{He-Li-V} designs are given in
\hbox{Table~\ref{t-6-3-1}}.

\clearpage

\begin{table}[h]
\caption{}
\centering{\bf Selected Input and Output Parameters \\
for the  He-Li-V and Ref-SEAFP Designs} \\
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l @{\hspace{0.5cm}} c c}
\hline
                                     &  He-Li-V     &  Ref-SEAFP  \\
\hline \\
Major Radius (m)                     & 6.75         & 9.4  \\
Minor Radius (m)                     & 1.5          & 2.09  \\
Fusion power (MW)                    & 1,925        & 3,000  \\
Average surface heat flux (MW/m$^2$) & 0.336        & 0.3--0.4  \\
Average neutron loading (MW/m$^2$)   & 2.9          & 2.1  \\
Structural material                  & V-alloy      & V-5Ti  \\
Helium pressure (MPa)                & 18           & 9  \\
Outlet helium temperature (\degC)    & 650          & 560  \\
Power conversion type                & Brayton-DCGT & Rankine  \\
Neutron multiplier                   & None         & None  \\
Tritium breeder                      & Li           & Li$_2$O  \\
Tritium breeding ratio               & 1.402 (1-D)  & 1.17 (3-D)  \\
Outboard module width (m)            & 1.086        & 1.46  \\
Inboard modules width (m)            & 1.021        & 1.44 \\
\\ \hline
\end{tabular}
\label{t-6-3-1}
\end{table}

\subsubsection{The He-Li-V blanket design}
\label{s-6-3-2-3}

To avoid the problems of compatibility between V-alloy and oxygen bearing
materials ({\it e.g.,} solid breeders), the option of using Li as the tritium
breeding material was reviewed.  This approach also eliminates all the design
difficulties related to the use of solid breeders, such as fabrication of
sphere-pac or sintered pellets, uniform filling of solid, loss of volume due to
void fraction, radiation damage, solid-to-solid contact resistance, sphere-pac
settling, thermal ratcheting effects, and the tritium purge flow system.  This
design also takes advantage of the high thermal conductivity of lithium,
thereby allowing relatively large spatial separation between cooling tubes.

\begin{figure}[p]
\centerline{\psfig{file=sti-6-3-04.ps,height=3.5truein}}
\caption[f-6-3-04]
{Cross section of a tokamak with a He-Li-V blanket.}
\label{f-6-3-04}
\vspace{1.cm}
\centerline{\psfig{file=sti-6-3-05.ps,height=3.5truein}}
\caption[f-6-3-05]
{Details of the He-Li-V blanket module.}
\label{f-6-3-05}
\end{figure}

The He-Li-V blanket is a toroidal flow, single wall tube design as illustrated
in \hbox{Figs.~\ref{f-6-3-04}} and \ref{f-6-3-05}. The Li is slowly circulated
for tritium extraction and, therefore, MHD insulation may not be necessary.
The blanket module consists of blanket boxes connected to the coolant plenum
and structural support.  As shown, the tube plates can be formed by embedding
(sandwiching) the coolant tubes in two V-alloy sheets ({\it e.g.,} by the HIP
process or by diffusive bonding).  This is similar to the ferritic steel solid
breeder design proposed by Kernforschungszentrum, Karlsruhe~\cite{Dal94}.
These plates form the support of the cooling tube and are also the supporting
structure of the breeder box, including the first wall.  Between the cooling
tube plates, the module is filled with molten lithium maintained at atmospheric
pressure, connected to a common channel, and then to the free surface of each
poloidal sector (not shown in \hbox{Fig.~\ref{f-6-3-05}}).  The coolant is
routed from the back poloidal plenum to cool the first wall and then re-routed
to cool the rest of the blanket.  This routing scheme is selected to optimize
the coolant outlet temperature in order to achieve the highest possible thermal
conversion efficiency.

A major-radius model neutronics calculation was performed for the He-V-Li
design.  The material volume fractions were determined after two iterations
between the configurational design, scoping thermal-hydraulics and neutronics
calculations.  Zone~1 (first wall and first layer of breeder) is 3~cm thick for
inboard and outboard blankets and Zone~2 (the rest of the blanket) are 28~cm
and 48~cm, respectively, for the inboard and outboard blankets.  The
composition of Zone~1 for both inboard and outboard is 7.3\% V-alloy, 79.2\%
lithium, and 13.5\% void.  The compositions of zone 2 are 7.0\% V-alloy, 91.6\%
lithium, and 1.4\% void. The results indicate that the 1-D tritium breeding
ratio is 1.402 and the total nuclear heating is 16.02 MeV per D-T neutron. A
detailed radial power density distribution for the outboard was used for the
scoping heat transfer calculation.

Based on the neutronics results, scoping calculations were performed to assess
the coolant and materials' maximum temperatures and the frictional pressure
drops for the blanket and first wall.  The primary and secondary stresses of
the first wall were estimated by the simple model of a circular tube.  Peak
surface and neutron wall loading assumptions were used for the maximum
temperature calculations.  \hbox{Table~\ref{t-6-3-2}} shows that the design can
meet temperature, structural design, and performance limits.

\begin{table}[p]
\caption{}
\centering{\bf The He-Li-V Thermal-Hydraulic Parameters$^{(a)}$}
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l @{\hspace{1.cm}} c}
\hline \\
Helium pressure (MPa)                               & 18.6  \\
Helium inlet temperature (\degC)                    & 400   \\
Helium outlet temperature (\degC)                   & 650   \\
Surface heat flux (MW/m$^2$)                        & 0.484 \\
\quad Peak-to-average ratio (outboard)              & 1.44  \\
\quad Peak-to-average ratio (inboard)               & 1.0   \\
Neutron wall loading (MW/m$^2$)                     & 4.83  \\
\quad Peak-to-average ratio (outboard)              & 1.67  \\
\quad Peak-to-average ratio (inboard)               & 1.0   \\
Inside diameter of first-wall tubes (mm)            & 10    \\
Inside diameter of blanket tubes (mm)               & 15    \\
Tube wall thickness (mm)                            & 2     \\
Blanket side wall thickness (mm)                    & 2     \\
First-wall temperature (\degC)                      & 611   \\
First-wall frictional pressure drop (kPa)           & 52    \\
Blanket front wall maximum temperature (\degC)      & 657   \\
Blanket front wall frictional pressure drop (kPa)   & 100   \\
First-wall primary stress (MPa)                     & 47    \\
First-wall total stress (primary + secondary) (MPa) & 94    \\
Gross thermal efficiency                            & 46\%  \\
\\ \hline
\multicolumn{2}{l}{$(a)$ See \hbox{Fig~\ref{f-6-3-05}} for other dimensions.}
\end{tabular}
\label{t-6-3-2}
\end{table}

Figure~\ref{f-6-3-06} shows a schematic of the blanket coupled with a direct
helium Brayton cycle for power production.  With an outlet temperature of
650\SdegC, a gross efficiency of 46\% can be attained.  Reasons for this high
value are: (1)~No intermediate heat exchange equipment is required, (2)~The
high pressure reduces blanket pressure losses, and (3)~A 96\% effective
recuperator with low pressure loss reduces cycle irreversibilities.

\begin{figure}[p]
\centerline{\psfig{file=sti-6-3-06.ps,height=3.3truein}}
\caption[f-6-3-06]
{Fusion Demo Plant Direct Helium Brayton Cycle Power Conversion.}
\label{f-6-3-06}
\vspace{1.cm}
\centerline{\psfig{file=sti-6-3-07.ps,height=3.4truein}}
\caption[f-6-3-07]
{Demo direct Brayton cycle efficiency as a function of maximum 
helium pressure and temperature.}
\label{f-6-3-07}
\end{figure}

One of the main advantages of the helium cooled blanket is its potential to
achieve much higher helium outlet temperature and hence higher thermal
efficiency.  Figure~\ref{f-6-3-07} shows the relationship between gross thermal
efficiency and temperature/pressure for a helium cooled blanket coupled to a
Brayton cycle.  A gross thermal efficiency of 60\% can be achieved with a
helium outlet temperature in the range of 1,100\SdegC.  This is well within the
capabilities of current gas turbines which operate at temperatures in the range
of 1,200--1,300\SdegC.  However, it will require an advanced blanket material
such as metal matrix composite or improved SiC/SiC composite.

It should be noted that in comparing the Rankine and Brayton cycles,
even when they have the same gross efficiency and pumping power, the net
thermal efficiency of the Brayton cycle system will be a few points higher
since the coolant pressure drop is already included in the DCGT system
calculation. 

Fig.~\ref{f-6-3-07} also illustrates the importance of gas pressure on cycle 
efficiency.  Above 12~MPa, performance becomes a relatively weak function of 
pressure.  Below that value, the pressure losses through the blanket become a 
dominant effect.  Therefore in the next design iteration, the coolant pressure 
will need to be optimized for blanket void fraction, gross efficiency, and 
pressure loss considerations.

\paragraph{Materials Issues}

Both helium and lithium are compatible with V-alloy through the entire proposed
temperature range of the \hbox{He-Li-V} design.  However, when helium is used
as the coolant with V-alloy, the interaction of impurities in He (O or H) with
V-alloy was identified as a critical issue in the BCSS study~\cite{BCSS}.
There has been no systematic program to address this issue.  Some experimental
data exist but mostly at relatively high concentrations of moisture
($\sim100$~ppm) content, atmospheric pressure oxygen, or air at high
temperature.  Therefore, it is important to review gas-cooled fission reactor
experience and identify the design differences between the fission and fusion
power plant in the control of impurity ingress.  Two additional approaches can
also be applied to mitigate this issue if the conventional method of impurity
extraction and control of impurities ingress proved to be inadequate.

{\bf Impurity removal.} The technology for impurity cleanup in helium coolant
loops has been well demonstrated in operating gas-cooled, fission reactors.
Experience in the U.S.\ and Europe has demonstrated systems with reliable
cleanup of impurity particulate, moisture, and gases using systems readily
available to the blanket cooling application.

The helium purification system takes a side stream of the primary coolant from
the discharge of the high pressure compressor discharge, provides essentially
100\% cleanup of the helium, and returns the purified coolant to the compressor
suction.  A typical helium purification train similar to what could be deployed
as a part of a blanket cooling system, consists of four factory-fabricated
process modules.  The modules and their functions are as follows:

\begin{itemize}

\item
\underline{High temperature filter/absorber}.  The first stage of the
purification process consists of an activated charcoal bed followed by a
metallic filter elements.  Condensible metallic impurities are removed from the
helium stream by absorption and chemisorption on the activated charcoal, acting
to filter out particulate.  Particulate ({\it i.e.,} dust) is mechanically
trapped by both the charcoal bed and metallic filter elements.

\item
\underline{Oxidizer}.  In the next purification stage, hydrogen, tritiated
hydrogen and carbon monoxide are oxidized to water and carbon dioxide for
subsequent removal.  Oxidation takes place at elevated temperatures as the
helium passes through an appropriate catalyst bed.

\item
\underline{Dryer}.  Water (including tritiated water) and carbon dioxide are
essentially completely removed at this stage by a molecular sieve.  In the
event of high tritium levels, separate water and carbon dioxide absorber
cartridges could be utilized.

\item
\underline{Low temperature absorber}.  The remaining gaseous impurities such as
nitrogen and any methane are removed as the He stream passes through a
liquid nitrogen-cooled, activated charcoal bed.

\end{itemize}

Systems processing more than 2,000~kg of helium per hour are well within 
existing experience base and would provide rapid coolant turnover, perhaps on 
the order of one to several hours, if so required. 

{\bf Control of impurity ingress.} In addition to cleanup, helium purity is
dependent on impurity ingress.  Precluding or tightly controlling impurity
ingress is requisite to maintaining coolant purity.  Experience with U.S.\ and
European helium-cooled reactors have demonstrated the practicality of achieving
low impurity levels.  \hbox{Table~\ref{t-6-3-3}} illustrates observed impurity
levels during reactor operations in the five helium-cooled reactors built to
date.  Impurity sources in these reactors included water from high pressure,
high temperature steam generators, impurities within graphite fuel and
reflector blocks, impurities trapped within fibrous insulating materials and
circulator bearing lubricants (oil or water).  Despite the presence of these
sources, low impurity levels were achievable in each of the systems.  In
particular, operating data from the German Thorium Hochtemperatur Reaktor
(THTR) (the most recently completed of these plants) shows oxidant (water)
levels to be less than their very low measurement threshold.

\begin{table}[t]
\caption{}
\centering{\bf Impurity Concentrations in Several Helium-Cooled Fission 
Reactors}
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l @{\hspace{2.cm}} c @{\hspace{1.cm}} c}
\hline
                      & H$_2$O  & H$_2$ \\
\hline \\
Dragon (U.K.)         & 0.1     & \Z1\D \\
Peach Bottom (U.S.)   & 0.5     &  10\D \\
AVR (Germany)         & 0.1     &  10\D \\
Fort St. Vrain (U.S.) & $<1$    & \Z2.6 \\
THTR (Germany)        & $<0.01$ & \Z0.8 \\
\\ \hline
\end{tabular}
\label{t-6-3-3}
\end{table}

In a He-V-Li blanket cooling system, each of these sources are eliminated.  
Steam generators are eliminated by adoption of the high efficiency Brayton 
Cycle (gas turbine).  This system is hermetically sealed to prevent any 
leakage during operation, but if a leak should occur, helium at a minimum of 
9~MPa would leak out rather than air leak in.  Molecular counter-diffusion of 
O$_2$ through the leak (low pressure) would be insignificant.  Likewise, 
leakage of water from the precooler or intercooler is precluded by the fact 
that the pressure of the helium (9~MPa) is far greater than the water 
(0.2--0.3~MPa).  Any leak would rapidly be detected by the presence of helium 
in the water.  Graphite blocks and fibrous insulation for the fission reactor 
are not required in a blanket cooling system.  This eliminates significant 
sources of helium impurities.  Finally, the maturation of magnetic bearing 
technology eliminates the potential for ingress of rotating machine 
lubricants.

It is likely that there will be oxide contaminated surfaces in the helium
system, especially the power conversion system which is constructed from high
alloy steels which develop oxide films.  This source of impurity will be
diminished significantly with the continuous operation of the purification
system.

Hence, the lack of ingress sources, coupled with a continuously operating
purification system, suggests that impurity levels very much lower (perhaps
several orders of magnitude) than those observed in helium-cooled reactors
might reasonably be expected.

{\bf Surface modification.} Since at this time there is no experimental 
data to demonstrate the equilibrium concentration of low level impurities in 
the helium primary loop of the He-V-Li blanket design and the sensitivity of 
V-alloy compatibility in this environment, it is reasonable to investigate 
various ways to mitigate the problem if warranted.

A possible way to make the V-alloy compatible with impurities is by way of
surface modification.  Surface modification to address V-alloy chemical
compatibility has been investigated by various scoping
studies~\cite{Nat95,Tob86,Cot72}.  In one case, aluminum was used to modify the
surface of \hbox{V-5Cr-5Ti} to make an Al rich surface region of the V-alloy.
This aluminized V-alloy had dramatically increased oxidation resistance at
temperatures to 575\SdegC in air when compared to standard \hbox{V-5Cr-5Ti}
material~\cite{Nat95}.  At 575\SdegC, the parabolic oxidation rate for
aluminized \hbox{V-5Cr-5Ti} was 1/7 the rate as that observed for
\hbox{V-4Cr-4Ti}.  In another case, chromium surface modification was used for
a \hbox{V-15Cr-5Ti} alloy to dramatically increase oxidation resistance,
relative to standard \hbox{V-15Cr-5Ti}, upon exposure to He containing 100~vppm
H$_2$O at temperatures to 650\SdegC~\cite{Tob86,Cot72}.  Chromium surface
modified V-15Cr-5Ti gained 2\% as much weight as \hbox{V-15Cr-5Ti} upon
exposure to He containing 100~vppm H$_2$O at 650\SdegC for 400 h.  Studies such
as these show the merit of surface modification of V-alloy structural material
to enhance chemical compatibility.

It can be envisioned that for surface modification, if necessary, 10--20~$\mu$m
layer of Si, Al or Y can be surface impregnated to the V-alloy surface facing
the high pressure helium.  Silicon and aluminum are semi-metals whereas yttrium
is a refractory metal. Silicon is an attractive element with regard to surface
modification of V-alloy material because of its high melting point and because,
once SiO$_2$ is formed on silicon from surface oxidation, the SiO$_2$ layer is
self-protecting.  Silicon oxide does not allow oxygen to penetrate and react
with underlying silicon metal.  In addition, silicon oxide is a fairly
impermeable barrier to hydrogen and does not react with hydrogen~\cite{Cot72}.

Aluminum will behave similarly to silicon if used as a protective surface
barrier for the V-alloy.  Aluminum is more ductile than silicon, however
aluminum by itself melts at 660\SdegC.  Yttrium is more of an unknown with
regard to use as a surface barrier to prevent chemical attack of V-alloy
material.  Yttrium does form very stable oxides that may be self-protecting for
underlying material as is the case with silicon.  It has already been shown
that V-alloy material can be produced successfully to have a 5--15~$\mu$m Al
rich surface layer~\cite{Nat95}.  Surface aluminized \hbox{V-5Cr-5Ti} prepared
by ANL showed a reduction in tensile ductility from 23\% to 13\% and no
catastrophic embrittlement of the alloy after exposure for 1000~h at 500\SdegC
in air.  This preliminary result is very encouraging for surface modified
V-alloy structural materials.

{\bf Oxidation resistant V-alloy.} Another approach to make the V-alloy
compatible with impurities is to develop an oxidation resistant V-alloy.
Researchers have already experimented with additions of up to 1\% by weight Si,
Al, and Y to bulk \hbox{V-5Cr-5Ti} and \hbox{V-15Cr-25Ti}
materials~\cite{Sat94}.  They have observed a degree of solution-strengthening
and enhanced oxidation resistance for these bulk modified V-alloys based on
measurements performed.  Oxide layers formed during exposure to pure O$_2$ at
400\SdegC were found to be cohesive during this work.

\paragraph{Design with High Pressure Helium.}

A system pressure of 18~MPa is relatively high but by no means unusual by
modern industrial standards.  For example, hundreds of superheated steam
systems operate in the range of 17--20~MPa and supercritical steam systems
operate in the range of 24~MPa.  Natural gas line pumping units operate at
pressures to 27.5~MPa.  

Two considerations are very important.  First, the incidence of small leaks
must be very low.  Second, if a small leak occurs, it must not rapidly
propagate to a large leak before action is taken. It is important to note that
the probability of a leak is a weak function of gas pressure.  Rather, it is
mainly a function of the incidence of undetected critical flaws and a crack
propagation mechanism such as creep or high or low cycle fatigue.  Stresses
from these mechanisms are much greater than primary (pressure induced)
stresses.  Current design practice, used in the utility and process industries,
employs established calculational procedures to determine critical flaw sizes
and inspection techniques to detect the presence of critical flaws.  This
approach has resulted in highly reliable operation of high pressure systems.

There are at least two methods to evaluate the probability of failure.  The
first method is used when we have neither the design nor the necessary
properties of the relevant material.  The failure rate is projected from the
operating experience of similar combinations of materials and design
conditions.  The usefulness of the result to a specific design depends very
much on the assumptions made in the extrapolation of data.  The second method
is to use projected irradiated properties of the relevant structural material
and to use fracture mechanics to project the result.  The accuracy of the
second method depends on the degree of conservatism in selecting the key
material properties.

T.~D.~Marshall~\cite{Mar94} used the first method to evaluate the failure rates
of the in-vessel ITER tubing made of different materials and coolants.
Operating experience failure rates and data for Nb and Ti were used to
represent Zr and V.  Failure data for Ti and water coolant and reliability
$K$-factors were used to make a projection for the vanadium and helium system.
An average failure rate of \hbox{$3.1\X 10^{-10}$~(m-hr)$^{-1}$} and an average
rupture failure rate of \hbox{$3.1\X 10^{-11}$~(m-hr)$^{-1}$} were found for
water, helium, and lithium as coolants and from Be to Zr as structural
materials including V-alloy.

A more specific evaluation based on the second method  was
performed by General Atomics on a similar helium-cooled, V-alloy design.  The
``leak-before-break'' assessment utilized by the utility and process industries
was used to evaluate the probability of rupture~\cite{Won93}.
This leak-before-break assessment should be applied to any pressurized
cooling media, including the lithium self-cooled option.  The approach is to
show that an incipient leak can be detected and terminated before a large
rupture occurs.  From knowledge of the baseline condition of the component and
type of failure mechanisms, material properties and anticipated operating loads
can be used to predict the distribution of times between crack initiation from
a given flaw size and leak and between leak and break.  This enables us to
design our in-service inspection procedures and safety instrumentation to
prevent ruptures from occurring in key components ({\it e.g.,} fissile reactor
primary coolant piping or vessels, and the divertor, blanket and first wall for
the fusion system).

In 1993, General Atomics applied the technique of leak-before-break assessment
on the helium-cooled blanket design for ITER~\cite{Won93}.  The material used
was V-alloy, the helium pressure was set at 20~MPa, and the helium-cooling
tubes were designed to 100,000 full power shots.  This large number of cycles
is a much more severe requirement than what is required for our Demo design,
which should be much less than 50 cycles for a 5-year life first wall.  In this
ITER blanket design exercise, conservative assumptions were made in the
calculations of crack propagation.  A value of $K_{IC}$ reduced by irradiation
to 40~MPa-m$^{1/2}$ was assumed for the critical crack length calculation.
(This is the irradiated lower shelf value for \HT.  It is a conservative value
for V-alloy because of its superior ductility at higher temperature.)  Analyses
were performed for different ITER tubes at different loading conditions.  The
surface loadings were for the cases of 0.35 and 1~MW/m$^2$.  Results showed
that an initial undetected flaw of less than 0.9~mm depth would not produce
leakage in less than 100,000 cycles ({\it i.e.,} each burn cycle equal to
1000~s).  This is a large initial flaw and can be assumed to have a very low
probability of occurrence because manufacturing techniques and quality-control
(QC) inspection would not allow such a flaw to exist in the base material.  In
addition, results also showed nearly all the failure scenarios met the
leak-before-break criteria.  Only in the case where the co-axial tube
inter-space was pressurized (20~MPa) and the loading was 1~MW/m$^2$ would
brittle failure result ({\it i.e.,} tube burst, fracture toughness exceeded).
No threshold $K_{I}$ was used, {\it i.e.}, crack growth was assumed to occur
even at very low stress levels.  These results are encouraging, especially when
applied to the less stringent Demo design, with considerable fewer cycles.  It
indicates that a robust design can be achieved by ensuring that both fracture
mechanics and ASME Boiler and Pressure Vessel code requirements are met.

Therefore, under normal operating conditions, the following general 
observations can be made:

\begin{enumerate}

\item
Helium will not leak unless there is a physical hole or crack; {\it i.e.}, no 
leakage can occur by diffusion.

\item
A cost-effective non-destructive inspection technique must be available ({\it
e.g.,} ultra sound or computerized tomography with further development to
reduce cost) to examine all the key pressure bearing components before, during,
and after installation.  (Such inspection techniques are available with the
capability of detecting axial and circumferential flaws of 0.1~mm in weld and
base metal tubes and components.)

\item
The leak-before-break condition must apply to both tubes and joints.

\item
The key to this design approach is to have reliable irradiated material data
(where $K_{IC}$ is the most important property) needed to carry out the
leak-before-break assessment.  Before this material data is available,
parametric studies can be performed to scope the problem for different design
options.

\end{enumerate}

{\bf Detection of helium leakage.} To ensure the application of the
leak-before-break technique and to avoid cooling tube bursting, it is necessary
to derive a detection technique for helium leakage for the \hbox{He-Li-V}
design.

At 2000~MW of fusion power, helium is generated in the lithium breeder at the
rate of 406~g/d, {\it i.e.}, about 0.0047~g/s.  Therefore, it may be difficult
to detect small leaks by monitoring the helium levels in the Li breeder.  One
solution is to add dopant to the primary coolant.  However, no doping materials
have been identified that will be reliable and will not add significant
radiation into the coolant stream.  From the technique of leak-before-break
scenarios, it may be possible to detect the amount of additional helium ingress
from the coolant stream without reaching the critical flaw size of about
0.9~mm.  Based on the observation that the helium leakage rate is proportional
to the flaw characteristic length squared, a flaw size in the range of 0.1 to
0.5~mm diameter (less than the critical flaw size of 0.9~mm) will lead to a
helium leakage rate in the range of 0.05 to 1.2~g/s, respectively.  This means
that helium leakage from the primary coolant can be detected based on the much
larger generation of helium from the leaking tube before the flaw size becomes
critical.  The response time for leak detection and termination will have to be
verified and coordinated with a specific blanket configuration and the response
of the passive fail-safe fast-activating valves.

{\bf Impact of tube rupture.} As part of the safety evaluation, even with 
the low probability of helium leakage and the fast response of leak detection 
without leading to any tube breakage, the scenario of tube failure still 
needs to be evaluated.  This is to show that, even with a tube 
failure in the blanket, the consequences are acceptable within anticipated low 
frequency of occurrence. 

Two approaches can be taken to design for the possibility of tube rupture.  The
first one is to design the blanket structure to withstand the full helium
pressure.  This approach is taken by the European DEMO BOT, ferritic steel,
solid breeder blanket design~\cite{Dal94}, operating at a gas pressure of
8~MPa.  For the \hbox{He-Li-V} design, an 18-MPa blanket module would have to
be configured to withstand this pressure for a duration shorter than or equal
to the response time for valving off the failed blanket module without
catastrophic investment impact.  With the higher strength and lower neutron
absorption of V-alloy over ferritic steel, this approach may be credible.  Note
that the selection of 18~MPa helium pressure is a scoping design point.
Various trade off studies will need to be performed between helium pressure and
performances of different areas, {\it e.g.}, thermal hydraulics, neutronics
design, structural design, and economics.

The second approach is to design the module with a plenum and relief mechanisms
to minimize damage (if any).  For this approach a free surface of liquid
lithium could be used and a dump tank of lithium could be envisioned to delay
the impact of module over-pressurization.  The viability of this approach may
be in question due to the same liquid metal MHD effect that is hindering the
lithium self-cooled design.  Very possibly some local damage can occur.  It
will still be tolerable provided the frequency of occurrence is low (0.001/y).
Based on the robust design approach, this frequency of occurrence should be
achievable.

Furthermore, we must show that a tube failure inside the blanket will not cause
major investment damage (damage that will cause major downtime of a year or
more).  The blanket module design consists of 1,760 tubes, with inner diameters
of 1.5~cm and length of 2.5~m.  The tubes are manifolded in two groups with
fail-safe isolation valves.  Passive isolation within 0.1--0.5 seconds should
be achievable according to instrumentation experts from General Atomics.  If a
tube rupture occurs, helium will flow into the module, causing distortion.  The
total helium leaked is determined by the leakage until the valve closes and the
equilibration of the volume in the tubes with the volume in the module.
Assuming a pressure of 8~MPa (taking the Ref.~\cite{Dal94} helium-cooled design
as an example) of the limit which would not cause catastrophic burst of the
module, the total volume of helium leaked into the module at 8 MPa is
$\sim0.5$~m$^3$.  Most of the leakage occurs from helium within the system
coming into equilibrium.  This means that the module would deform by 0.5 cubic
meters.  This volume deformation is about 11\% of the total volume of the 1~m
wide, 0.5~m deep, and 8~m high module.  We have to design for the lithium
displacement of this volume before a configuration change of the module would
occur.  This requirement does not seem to be excessive, but it will have to be
demonstrated.

{\bf Consequence of helium release into the confinement building.} For complete
safety evaluation, the potential consequences of a complete loss of helium
volume should be analyzed.  An estimate can be made by scaling the results from
the ITER design~\cite{Won94}.  For this ITER design, preliminary safety
assessments of various pipe failure conditions were performed.  At a system
helium pressure of 19~MPa, with a volume of 150~m$^3$, $T_{in}$ at 35\SdegC,
and $T_{out}$ at 350\SdegC, the energy content is 4.2~GJ.  The helium volume at
atmospheric pressure is 28,000~m$^3$.  Assuming an ITER crane hall volume of
150,000~m$^3$, the pressure in this hall after release of helium is $<19$~kPa
(without any relief), and the confinement building would be at $<35$\SdegC.

Based on the above, we can make estimates for the Demo design as follows.  The
blanket He volume is $\sim120$~m$^3$.  External systems are included by
doubling the inside volume to 240~m$^3$.  With a 150,000~m$^3$ confinement hall
volume~\cite{Won94}, the pressure on the wall will be $<31$~kPa.  The energy
content is $\sim5.3$~GJ.  Therefore, the confinement building temperature
following a complete loss of coolant will be $\sim39$\degC.  If the entire
helium coolant is leaked into the vacuum chamber (with a volume of 3,300
m$^3$), the helium pressure would be 1.4~MPa which is large but not excessive.
To mitigate the potential damage, a passive pressure release valve can be
designed into the vacuum system if necessary.

{\bf Maturity of helium-cooling technology.} Commercial fission gas-cooled
reactors have been operating in Europe and in the U.S.\ for more than
1,000~reactor-years (primarily CO$_2$), providing experience with gas
circulators and heat exchanges.  In addition, five helium Brayton cycle systems
have been constructed and operated.  Developmental helium-cooled reactors like
Peach Bottom were successfully demonstrated and a commercial reactor has been
operated.  High temperature helium test loop experience has been obtained in
facilities in the U.S., Japan, Germany, Russia and France.  These indicate the
technology of helium-cooling is relatively matured.  Even though the
development of gas-cooled commercial fission reactors has been relatively slow,
considerable advances have been made in the last 20~years.  This can be
understood from the development of the passively-safe Modular High Temperature
Gas-cooled Reactor (MHTGR) and the Director Cycle Gas Turbine GT-HTGR reactor
designs.  For the development of a gas-cooled Demo design, we should be able to
take advantage of the significant saving we can realize by making use of the
existing technology of helium gas cooling, which is a relatively matured power
plant technology.

\subsubsection{Divertor and in-vessel component options}
\label{s-6-3-2-4}

As an engineering system, the in-vessel components, including the divertor
design requirements are: (1)~to handle the high average and maximum surface
heat flux while optimizing the coolant outlet temperature and minimizing the
pumping power, (2)~to use a surface material which minimizes the sputtering
erosion and maximizes the component life-time, (3)~to use a sound and simple
mechanical design to provide high reliability and allow ease of maintenance,
(4)~to withstand a few disruptions, and (5)~to have a vacuum pumping system to
match the requirements of ash removal and density control of the divertor and
impurity control of the plasma-core and divertor channels/plates.  In the
following, we evaluate the feasibility of using V-alloy to remove the maximum
heat flux of 5~MW/m$^2$ for both lithium and helium coolants and review the
status of using V-alloy as the plasma facing material.

\paragraph{Thermal-hydraulic evaluation.}

As a scoping feasibility study, only a simple 2-m long tube with 2-mm coating
of W was used for this evaluation.  The cooling tube was used to remove the
impinging average surface heat flux of 1~MW/m$^2$ and maximum surface heat flux
of 5 MW/m$^2$. Only surface heat fluxes impinging onto the divertor surfaces
were considered (the contribution of the volumetric power generation would be
less than 10\% of the surface heat flux).  The maximum V-alloy design
temperature limit of 700\SdegC and the primary and primary plus secondary
stress limits of $S_{mt}=105$~MPa and $3S_{mt}=315$~MPa, respectively, at
fluence of 200~dpa, were used as our design guidelines~\cite{BCSS}.

Adjustable parameters are the tube diameter, wall thickness, and the coolant
inlet and outlet temperatures.  The coolant flow rate was defined by the
selected coolant inlet and outlet temperatures and the average power
intercepted by the projected surface area of each coolant tube.  With the
knowledge of coolant volume flow rate and tube inside diameter, the coolant
velocity, heat transfer coefficients, tube wall, and sacrificial wall
temperature gradients were calculated.  The maximum tube temperature was then
determined as the sum of the bulk coolant temperature, the film drop
temperature, and solid material temperature gradients.  These calculations were
applied to both coolant options.

The first option is the lithium-cooled design.  The lithium coolant operates in
a magnetic field strength of 7~T, with a tube diameter of 1~cm.  The Hartmann
number of the design is 4,702, which is well into the laminar regime indicated
by Gardner~\cite{Gar68}.  Therefore a Nusselt number of $Nu=7$ was used for
the calculation.  This value of $Nu$ was determined from laminar pipe flow in a
transverse magnetic field with uniform wall heat flux; therefore, to compensate
for the one-sided heat transfer, a flow diversion design such as helical fins
may need to be incorporated in the future design.  This local flow distribution
will only be needed at the location of maximum heat flux.  Here we also assumed
that the necessary MHD insulation inside all the coolant channels will function
properly and will not impact either the heat transfer or the pressure drop of
the design.

The second option is the helium-cooled design.  In order to enhance heat
transfer, surface roughening was assumed.  This allows an improvement of heat
transfer by a factor of two with the corresponding friction factor increased by
a factor of 3~\cite{Dal79}.  Similar to the lithium design, a twisted tape also
can be used to reduce the effect from one-sided heat flux~\cite{Kak87}.  Again
only local heat transfer enhancement will be needed at the location of the
maximum surface heat flux.

Table~\ref{t-6-3-4} shows the scoping design results of the lithium and
helium-cooled design options.  These results show that, to meet the thermal
stress limit, the V-alloy tube wall will have to be relatively thin.
Correspondingly, relatively small diameter tubes will be necessary to generate
high enough heat removal capacity to meet the material temperature limit.
Tubes with inside diameters of 1 and 0.6~cm can be used for the lithium and
helium coolant options, respectively.  The coolant outlet temperatures also
will have to be lower than the corresponding blanket outlet temperatures to
meet the maximum temperature limit of 700\SdegC for the V-alloy.  Relatively,
the helium-cooled tubes will have to be smaller and will require higher
pumping power than the lithium-cooled design.

\begin{table}[tb]
\caption{}
\centering{\bf Scoping Thermal-hydraulics of V-Alloy Divertors}
\vspace{1.0cm}
\renewcommand{\arraystretch} {1.3}
\begin{tabular}{l @{\hspace{0.5cm}} c c}
\hline \\
Coolant                               & Lithium     &  18~MPa He  \\
Inlet temperature (\degC)             & 300         &  250  \\
Outlet temperature (\degC)            & 450         &  400  \\
Gross thermal efficiency              & $\sim39\%$  &  $\sim39\%$ \\
Tube diameter (cm)                    & 1           &  0.6  \\
Tube-wall thickness (mm)              & 1           &  1  \\
Coolant velocity (m/s)                & 1.3         &  78 \\
Heat-transfer coefficient, $h$ (W/m$^2$\degC)
                            & $3.3\X 10^4$ $^{(a)}$ &  $2.5\X 10^4$ $^{(b)}$\\
Tube-wall maximum temperature (\degC) & 694         &  690  \\
W-coating maximum temperature (\degC) & 793         &  790  \\
Pumping power fraction                & 0.01        &  0.09$^{(c)}$  \\
Primary stress (MPa)                  & 13          &  54  \\
Thermal stress (MPa)                  & 157         &  157  \\
Total stress (MPa)                    & 170         &  211  \\
\\ \hline
\multicolumn{3}{l}{$(a)$ A Nusselt number value of 7 was used.}  \\
\multicolumn{3}{l}{$(b)$ With local heat transfer enhancement of 2.}  \\
\multicolumn{3}{l}{$(c)$ With increased friction factor of 3.}  \\
\end{tabular}
\label{t-6-3-4}
\end{table}

\paragraph{Sputtering erosion and plasma facing material selection.}

Material erosion is an important consideration of in-vessel and divertor
component design and affects the life-time of the components.  Unfortunately,
for the V-alloy design, there is no relevant erosion data.  It is also possible
that V-alloy structure must be coated with another material such as C, Be, and
W. Carbon has been used as the first wall material for all existing large
tokamaks, like JET, JT-60, TFTR, and DIII-D.  Beryllium has been used in JET.
Carbon, Be, and W are the reference plasma facing materials as reported in the
fourth report for the Technical Advisory Committee (TAC) of the ITER-EDA
design~\cite{ITER-TAC4}.  Comparing the materials of interest, a potential
sputtering rate can be listed in the order of Be, C, V-alloy and W, where Be
will have the highest erosion rate and W the lowest~\cite{Bro94}. Highly
preliminary results based on present physics and modeling assumptions indicate
2 to 3 orders of magnitude difference between Be (at tenths of a meter per full
power year) and W in net erosion~\cite{ITER-DIV}. Due to its chemical
reactivity with hydrogen, C is potentially the worst when chemical sputtering
at high surface temperature is considered.  Under high fusion neutron fluence,
both Be and C have the additional disadvantage of significant loss in thermal
conductivity and structural strength.  Therefore both Be and C are projected to
be unsuitable as structural and coating material for the Demo design.  For
V-alloy, experimental measurements from a metal-coated sample exposed to the
DIII-D plasma indicated that the erosion of V can be at least an order of
magnitude lower than C, possibly not as low as W~\cite{Wam95}. While W exhibits
the best properties against erosion, contamination of plasma with W as well as
induced radioactivity in W are areas of concern.

\subsubsection{Use of vanadium in the shield}
\label{s-6-3-2-5}

Vanadium possesses salient features that are of great importance for fusion
plants.  Among these features are the high operating temperature capability
($\sim700$\SdegC), low afterheat generation, and low activation
characteristics.  However, V is a more expensive and less efficient shielding
material compared to steels.  Recent studies have indicated that, in order to
improve the economics of V-based fusion plants, it is essential to limit the
use of V to components where these features are most needed, {\it i.e.}, in the
plasma facing components, blanket, reflector, and inner layer of the shield.
Furthermore, less expensive, low activation materials (such as ferritic steels)
could be used in the outer layer of the shield and the vacuum vessel where
radiation flux and operating temperature are much reduced.

The economic analysis of the ARIES-II design, which employs V structure for the
entire shield, has indicated that the shield is the most expensive component in
the machine.  The shield is well optimized for high neutronics performance and
contains 15\% V structure, by volume.  Nevertheless, V comprises more than half
of the shield cost.  This has prompted the need to replace the V structure in
the outer layer of the shield with less expensive steel structure in order to
reduce the cost of future V-based designs.  This change should not impact the
safety characteristics of the design as the outer layer of the shield is
subjected to low radiation flux, generating low levels of radioactivity and
afterheat.  The saving in the shield cost will be significant ($\sim40\%$).
This will come at some cost in the power balance since steel cannot operate at
temperatures as high as V.  The outer layer of the shield and the vacuum vessel
will thus operate at lower temperature ($<$550\SdegC) and the small amount of
nuclear heating deposited therein will not be recovered in the power cycle.
The dividing boundary between the two layers of the shield will therefore
depend on how much power could be dumped as low grade heat with minor effect on
the power balance ({\it e.g.,} 1\%--5\%).

To conclude, it may not be economical to use V structure in the entire shield.
The inner layer of the shield, along with the first wall, blanket, and
reflector, should employ V structure whereas the lower temperature outer layer
of the shield and vacuum vessel should utilize stainless steel as the main
structural material.  This approach for designing the shield of V-based fusion
power plants, along with other innovative ideas to improve the shield
performance, will have significant impact on the cost of the shield and overall
machine.

\subsubsection{Use of vanadium in the heat transport loop}
\label{s-6-3-2-6}

For use in the blanket, the cost of the vanadium can be offset by the
performance of the material.  For the heat transport loop design, the use of
V-alloy has to be minimized to reduce the capital cost of the power plant.
Stabilized ferritic steel is a possible material to be used for the primary
loop.  However, corrosion of ferritic steel with Li at the coolant temperature
of ARIES designs can be severe.  Also, mass transfer between the steel loop and
the vanadium blanket may cause bi-metallic loop material problems.

The design proposed here uses double-walled tubing for the primary loop.  A
thick steel layer will be placed on the outside of the tube for support.  A
vanadium liner, about 1~mm thick, can be the inside layer facing the lithium.
The intermediate heat exchanger (IHX) design also can be a double-walled
design, which is similar to the ones developed for the breeder
program~\cite{Grexx}.  A sketch of the proposed IHX is shown on
\hbox{Fig.~\ref{f-6-3-08}}.  The primary coolant, lithium, will be on the tube
side, which will be connected to the outer tube sheet.  The outer tube of the
IHX can be made of stainless steel and will face the secondary coolant, sodium.
The outer tube will be connected to the inner tube sheet.  The joint between
vanadium and the steel can be made in the middle of the outer tube sheet.
Thus, the primary coolant, lithium, will never contact the steel, while the
secondary coolant, sodium, will never contact the vanadium.  The potential
problems associated with bi-metallic loop can thus be avoided.  To fabricate
this double-wall design, the vanadium will need to be welded to steel.
This has been demonstrated on a small scale.

The total volume of vanadium being used in the blanket and shield is 80~m$^3$,
while the volume used in the primary loop is estimated to be $\sim13$~m$^3$(
$\sim0.6$~m$^3$ for the main coolant pipes, $\sim10$~m$^3$ for the IHX tubes
and $\sim2$~m$^3$ for the IHX tube sheets, see \hbox{Fig.~\ref{f-6-3-08}}).
Assuming a unit cost of \$300/kg, the additional cost of vanadium used in the
heat transport loop is 24~M\$.

\begin{figure}[tb]
%\vspace*{4.2in}
\centerline{\psfig{file=sti-6-3-08.ps,height=3.25truein}}
\caption[f-6-3-08]
{A proposed IHX design for self-cooled Li/W designs.}
\label{f-6-3-08}
\end{figure}

\subsection{Material Properties}
\label{s-6-3-3}

Several of the properties which are most important in determining the 
attractiveness of vanadium alloys as fusion in-vessel materials are discussed 
in this section.  These include tensile and creep properties, irradiation
effects (swelling, creep-enhancement, and hardening or softening),
compatibility with gaseous and Li environments, and fabrication concerns.

A more comprehensive collection of properties is included as 
an appendix to this report.  These cover a wider range of properties that are
needed to complete the design phase of this effort.

The U.S.~Fusion Materials Program has identified V-4wt.\%Cr-4wt.\%Ti
(\hbox{V-4Cr-4Ti}) with 0.01 to 0.1~wt.\% Si as a promising structural material
for fusion reactor applications.  Recent progress made on this material is
summarized in the U.S.\ Contribution 1994 Summary Report on Task
T12~\cite{Smi94} from which most of the following material is taken.  One
laboratory heat (30~kg) and one industrial-size heat (500~kg) of this alloy
have been produced for laboratory-scale and in-reactor testing.  In addition to
alloying elements (V, Cr, Ti, Si) and impurities (H, O, C, N, {\it etc.}), heat
treatment plays a very important role in the performance of V-alloys.

The data base for binary (V-Ti, V-Cr) and ternary (\hbox{V-Cr-Ti}) alloys is
much more extensive than the data base for \hbox{V-4Cr-4Ti} alone.  This
extensive data base has been used to select the current reference alloy and,
for some properties ({\it e.g.,} thermophysical, elastic, {\it etc.}), it can
be used directly for the reference properties.  However, because heat treatment
and impurities can be as important as the base alloying elements, it has
sometimes been difficult to isolate the effects of each alloying element.  For
example, the excellent ductility of the \hbox{V-4Cr-4Ti} alloy as compared to a
particular heat of \hbox{V-5Cr-5Ti} led some researchers to speculate that the
properties are extremely sensitive to the sum of Ti and Cr content.  Subsequent
work in which the same purity of raw materials and the same heat treatment used
to produce the \hbox{V-4Cr-4Ti} alloy was used to produce the \hbox{V-5Cr-5Ti}
alloy resulted in similar properties for the two alloys.  Also, the
``optimization'' which has led to the selection of \hbox{V-4Cr-4Ti} has been
performed from a materials perspective, rather than from a design perspective.
Therefore, as more experience is gained with alloys in this class and as more
interaction takes place among the design engineers and the material scientists,
chemistry and heat-treatment modifications will be made to the reference alloy
to optimize V-Cr-Ti for design application

The \hbox{V-4Cr-4Ti} alloy, along with a specific heat treatment, has been
chosen as the reference alloy based on an optimization with regard to tensile
and impact ductility.  The addition of Si, along with limiting the interstitial
oxygen, carbon and nitrogen impurities to $\leq1,000$~wppm, is important for
swelling resistance.  While not necessarily optimized in the reference
composition, alloys with $\geq3$~wt.\% Ti and about 0.5~wt.\% Si promote
swelling resistance through the formation of finely dispersed Ti$_5$Si$_3$
precipitates.  With regard to mechanical properties, the tensile strength tends
to increase with increases in Ti and Cr content, as well as Si content.
However, the ductile-to-brittle transition temperature (DBTT) also increases
with these additions.  From a DBTT standpoint, it appears to be desirable to
keep the sum of the Ti and Cr concentrations to $\leq10$~wt.\%.  Similarly,
small increases in Si have a positive effect on strength, but tend to raise the
DBTT.  Historically, the emphasis on tensile and impact ductility comes from
the experience with ferritic steels which have problems in these areas due to
irradiation-induced hardening at the lower temperatures ($\leq400$\SdegC).
However, there is room for improvement by optimizing both the chemistry and
heat treatment from the perspective of design performance.  For example,
optimizing the alloy composition based on maintaining a DBTT of $\sim200$\SdegC
at the expense of some other property ({\it e.g.,} tensile strength, creep
strength, swelling resistance) makes little sense from a design perspective,
which at most requires a DBTT below room temperature.  Optimizing the alloy
composition to attain the highest uniform elongation may make little design
sense because all that is required of this parameter is it be $\geq5\%$ to
qualify as a ductile material.  Optimizing the alloy composition to improve the
yield strength would be of no direct benefit, because the ultimate tensile
strength ($S_u$) of \hbox{V-4Cr-4Ti} determines the allowable primary membrane
stress ($S_m=S_u/3$) and the combined membrane plus bending plus thermal stress
($S_u$).

\subsubsection{Thermophysical and elastic properties}
\label{s-6-3-3-1}

The density, heat capacity, thermal conductivity, Young's modulus and shear
modulus of pure vanadium are all well characterized as functions of
temperature.  Alloying elements Ti and Cr in the combined range of
$\leq10$~wt.\% have no significant effect on these properties.  Similarly,
non-metallic impurities in the reference range of $\leq1,000$~wppm have no
effect.  Neutron damage to the bulk material should also have no significant
effect.  As compared to austenitic and ferritic steels, the \hbox{V-4Cr-4Ti}
alloy is lighter (6.07~g/cm$^3$ at room temperature), higher in thermal
conductivity, and lower in thermal expansion.  When coupled with its higher
yield and ultimate strength at high temperature, \hbox{V-4Cr-4Ti} is a better
high heat-flux material than either of the steels.

\subsubsection{Tensile and creep properties of unirradiated \hbox{V-4Cr-4Ti}}
\label{s-6-3-3-2}

The ultimate tensile strength ($S_u$), tensile yield strength ($S_y$), uniform
elongation ($e_u$) and total elongation ($e_t$) of \hbox{V-4Cr-4Ti} are
reasonably well characterized from room temperature to 700\SdegC.  More data on
the industrial heat are forthcoming to add some statistical significance to the
overall data base.  Data are also being generated on the reduction in area (RA)
at the failed neck region, which is a better measure of local failure strain
than $e_t$.  The secondary thermal creep rate and time-to-rupture of
\hbox{V-4Cr-4Ti} are characterized at 600\SdegC for high stresses
(380--420~MPa).  Higher temperature and lower stress data are needed.

Based on the tensile and creep properties, \hbox{V-4Cr-4Ti} is a better
high-temperature structural material than either austenitic stainless steel
({\it e.g.,} 316L(N)) or ferritic steel ({\it e.g.,} \hbox{Fe-9Cr-1MoVb}).  Its
ultimate tensile and yield strengths are relatively constant for temperatures
in the range of room temperature to 700\SdegC.  In this section, the properties
of \hbox{V-4Cr-4Ti} are compared to those of 316L(N) and \hbox{Fe-9Cr-1MoVNb}
in terms of two parameters which are of particular interest to the design of
first-wall and high heat flux components.  The first parameter is the maximum
allowable primary membrane stress ($S_m$).  In general, $S_m$ is defined as the
minimum of $2~S_{yd}/3$ and $S_{ud}/3$.  For austenitic stainless steels,
values of $S_m$ as high as $0.9S_{yd}$ are allowed under certain circumstances
({\it e.g.,} $T\geq160$\SdegC).  The parameters $S_{yd}$ and $S_{ud}$ are
``design'' values for the yield and ultimate tensile strengths, respectively.
The design properties are determined from the data base by bounding from below
the room temperature strengths and then following the temperature dependence
determined from the average of the data.  In a statistical sense, the
properties are neither ``average'' nor ``lower-bound'' properties over the
temperature ranges of interest.  However, they are consistent with what is
required by design codes for determining maximum allowable stresses.  For both
316L(N) and \hbox{Fe-9Cr-1MoVNb}, the values are taken from the French RCC-MR
design code~\cite{Tav95,War95}. In the case of \hbox{V-4Cr-4Ti}, which is not
as yet code-qualified and for which there is only a limited data base, the
scaling factors for 316L(N) are used: $S_{yd}=S_{y}(avg.)/1.25$ and
$S_{ud}=S_{u}(avg)/1.15$.

The results of the S$_m$ comparison are shown in \hbox{Fig.~\ref{f-6-3-09}}.
Above 350\SdegC, \hbox{V-4Cr-4Ti} is a stronger material than 316L(N) against
primary loads ({\it e.g.,} coolant pressure).  Above 525\SdegC,
\hbox{V-4Cr-4Ti} is stronger than \hbox{Fe-9Cr-1MoVNb}.  These results do no
take into account the long term effects of creep and fatigue.

\begin{figure}[p]
\centerline{\psfig{file=sti-6-3-09.ps,height=3.05truein}}
\caption[f-6-3-09]
{Comparison of the allowable primary membrane stress (S$_m$) for unirradiated
316L(N) stainless steel (based on RCC-MR code), ferritic/martensitic
\hbox{Fe-9Cr-1MoVNb} (based on RCC-MR code) and \hbox{V-4Cr-4Ti} (based on 1/3
of the minimum ultimate tensile strength, which is taken as the average/1.15).}
\label{f-6-3-09}
\vspace{1.cm}
\centerline{\psfig{file=sti-6-3-10.ps,height=3.05truein}}
\caption[f-6-3-10]
{Comparison of the thermal stress factor for 316L(N),
\hbox{Fe-9Cr-1MoVNb}, and \hbox{V-4Cr-4Ti}.}
\label{f-6-3-10}
\end{figure}

The second parameter of interest in the design of first-wall and high-heat-flux
components is the thermal stress factor (TSF).  The sum of primary pressure and
bending loads, along with secondary stresses such as thermal stresses, is
limited to $3S_m$, which for Fe-9Cr-1MoVNb and \hbox{V-4Cr-4Ti} is equal to
$S_{ud}$.  The thermal stress (or maximum heat flux) factor is a measure of the
maximum surface heat flux (in MW/m$^2$) allowed for a 1-mm-thick wall.  It is
calculated from the following formula which is based on the maximum allowable
stress (3S$_m$) divided by an estimate of the thermal stress:
%
\begin{eqnarray}
{\rm TSF} & = & \frac{3 S_m}{(\alpha_m E)/[k (1-\nu))]} \: ,
\label{e-6-3-1}
\end{eqnarray}
%
where $\alpha_m$ is the mean coefficient of linear thermal expansion (in
10$^6$/\SdegC), $E$ is the Young's modulus (in GPa), $k$ is the thermal
conductivity (in W/m-\degC) and $\nu$ is Poisson's ratio.  The results are
shown in \hbox{Fig.~\ref{f-6-3-10}} for the three materials of interest.  These
results can be interpreted in the following manner.  At 650\SdegC, the thermal
stress factor for both 316L(N) and \hbox{Fe-9Cr-1MoVNb} is about 1.8, while for
\hbox{V-4Cr-4Ti} the factor is about 6.0.  For a 5-mm first wall thickness at
650\SdegC, \hbox{V-4Cr-4Ti} can accommodate a surface heat flux as high as
1.2~MW/m$^2$, while the other two materials can only accommodate a surface heat
flux of 0.36~MW/m$^2$.  The results for both \hbox{V-4Cr-4Ti} and 316L(N) are
relatively insensitive to temperature, while \hbox{Fe-9Cr-1MoVNb} performs much
better at lower temperatures.  These calculations assume no primary loads on
the structure.  A more detailed calculation would start with an assumed local
coolant temperature and use iteration until a consistent set of
temperature-dependent material properties and limiting heat flux were achieved
for a given wall thickness.

\begin{figure}[b]
\centerline{\psfig{file=sti-6-3-11.ps,height=3.2truein}}
\caption[f-6-3-11]
{Secondary thermal creep rates of 316L(N), \hbox{Fe-9Cr-1MoVNb}, and
\hbox{V-4Cr-4Ti} at 600\SdegC.  The correlation plotted for
\hbox{Fe-9Cr-1MoVNb} is actually for HT-9 (\hbox{Fe-12Cr-1MoVW}) which should
represent an upper bound to \hbox{Fe-9Cr-1MoVNb}.}
\label{f-6-3-11}
\end{figure}

In the previous comparisons among \hbox{V-4Cr-4Ti}, 316L(N) and
\hbox{Fe-9Cr-1MoVNb}, the $S_m$ limit to primary membrane stresses and the
$3S_m$ limit to combined primary membrane, primary bending and thermal stresses
were used.  These limiting parameters ensure reasonable design margins
excluding the possible time-dependent effects of creep and fatigue.  Creep is
incorporated into the analysis by making $S_m$ time-dependent ($S_{mt}$) based
on a limiting creep strain or time-to-rupture and a desired lifetime.  Thus,
the higher the creep-resistance of the structural material, the higher the
$S_{mt}$ values.  While creep data for \hbox{V-Cr-Ti} ternary alloys are
available, the thermal creep data base for \hbox{V-4Cr-4Ti} is limited a few
high-stress data points at 600\SdegC.  Thus, the secondary thermal creep rates
of the three structural materials are compared in \hbox{Fig.~\ref{f-6-3-11}} at
600\SdegC.  The \hbox{V-4Cr-4Ti} ally is clearly the most creep-resistant at
600\SdegC.  In terms of the reduction in $S_m$ with time ($S_{mt}$) over a
two-year period, thermal creep tends to cause a reduction for
\hbox{Fe-9Cr-1MoVNb} above 450\SdegC, for 316L(N) above 550\SdegC, and for
\hbox{V-4Cr-4Ti} above 650\SdegC.  For a four-year time period corresponding to
a lifetime of 200~dpa, it does not appear likely that thermal creep will cause
a reduction in $S_m$ ($\sim125$~MPa) for temperatures $\leq650$\SdegC.
Clearly, more thermal creep data for the reference alloy are needed to resolve
this issue.

\subsubsection{Irradiation effects}
\label{s-6-3-3-3}

Irradiation effects of importance to structural materials in a fusion
environment include swelling, creep-enhancement, hardening (or softening) and
embrittlement.  Swelling data on V-alloys have been obtained for temperatures
in the range of 420 to 600\SdegC up to 114~dpa.  Unlike the behavior of 316L(N)
and \hbox{Fe-9Cr-1MoVNb}, V-alloy swelling tends to peak in the neutron damage
range of 30--80~dpa and then to decrease.  The peak volumetric swelling
measured for \hbox{V-15Cr-5Ti} is in the range of 2\%--3\% at 114~dpa.
Generally, a 5\% design limit is placed on volumetric swelling.  However,
deformation limits which do not lead to failure of a component in isolation are
very design dependent.  In the spirit of this generalization, swelling of
V-alloys should not limit the lifetime of components for the dpa range of the
data base ($\leq114$~dpa).  While there is no analytical or experimental
evidence to suggest that volumetric swelling would limit operation to higher
damage levels ({\it e.g.,} 200~dpa), this needs to be demonstrated for the
reference alloy for He/dpa ratios of $\leq6.5$.

The swelling of the reference V-alloy (\hbox{V-4Cr-4Ti}) has been investigated
in the temperature range of 425--600\SdegC up to 31~dpa with He/dpa values
ranging from $\ll1-4.2$.  The peak swelling observed is 1.8~vol.\% at
425\SdegC, 31~dpa and 1~appm He.  Assuming that this alloy continues to
perform the same as the other V-alloys tested, then peak swelling values in the
temperature range of 400--600\SdegC should remain below 5~vol.\% beyond
100~dpa.

Irradiation-enhanced creep has been studied for austenitic and ferritic steels,
but not for V-alloys.  Unlike thermally-induced creep, this deformation
mechanism is considered to be ``non-damaging'' in the sense that there is no
ductility limit associated with, or correlated with, failure.  In a sense, it
is similar to swelling in that large amounts of strain could cause geometric
problems and differential strains across the component could result in
significant secondary stresses.  It is commonly assumed that the irradiation
creep rate of V-alloys will be similar to that of austenitic and ferritic
steels.  Because of the complication of swelling in the austenitic steel data
base, the irradiation creep correlation for low-swelling ferritic steel is
usually used to represent the behavior of V-alloys.  An approximation for the
creep strain for both 12Cr and 9Cr ferritic steel is~\cite{Gar91}:
%
\begin{eqnarray}
 \epsilon_{ic} &  = & 2.2\X 10^2 \, D \, \sigma \: ,
\label{e-6-3-2}
\end{eqnarray}
%
where $\epsilon_{ic}$ is the irradiation creep, $D$ is neutron damage in dpa
and $\sigma$ is effective stress in MPa.

For \hbox{V-4Cr-4Ti}, $S_m=125$~MPa for 500--700\SdegC.  Using a 5\% limit for
irradiation creep means that at 125~MPa the 5\% limit would be reached at
180~dpa.  In order to achieve a lifetime of 200 dpa at $\leq5\%$ irradiation
creep strain, $S_m$ would have to be reduced by 10\% to 113~MPa.  However, in
future work, the irradiation-induced creep rate of \hbox{V-4Cr-4Ti} needs to be
determined directly and, once a specific design has been established, an
irradiation creep strain limit needs to be established based on the design
details.

Thus far in the analysis, it appears that neither swelling nor irradiation
creep of \hbox{V-4Cr-4Ti} would limit operation of \hbox{V-4Cr-4Ti} to
$<100$~dpa.  Operation to damage levels up to 200~dpa may be achievable, but
data are required to substantiate safe, reliable performance up to such high
damage levels.  

The next issues addressed are hardening and embrittlement.  Within the
irradiation data base available for \hbox{V-4Cr-4Ti} (400--600\SdegC,
0--31~dpa, and 0--75~appm He), the changes in tensile properties are as
follows: UTS increases by 79\% (400--430\SdegC), 35\% (500--520\SdegC) and 8\%
(600\SdegC); YS increases by 162\% (400--430\SdegC), 112\% (500--520\SdegC),
and 45\% (600\SdegC); UE decreases to $1.6\%\pm1.0\%$ (400--430\SdegC),
$3.6\%\pm1.3\%$ (500--520\SdegC) and $6.5\%\pm1.1\%$ (600\SdegC).  For all of
these cases, the DBTT remains $\ll$0\SdegC.  The increase in strength should
cause no problems in the design.  However, the designer should not use this
irradiation-strengthening to increase the allowable stresses.  Overpower and
undercooling events which tend to elevate the structure temperature for brief
periods of time may anneal out this hardening.  Also, with some materials [{\it
e.g.,} 316L(N)] cycling has been observed to soften the irradiated material.

The decrease in uniform elongation with irradiation is significant for samples
irradiated and tested at 400--430\SdegC.  Such low uniform elongation values
require the use of special design rules such as those developed for ferritic
and austenitic steels which tend to embrittle to even lower UE values at
200--400\SdegC.  They also suggest that more irradiation testing be performed
to map out the temperature range, neutron damage level and He concentration at
which embrittlement may become a serious design issue.  In the interim, it is
recommended that T > 430\SdegC be used as a lower limit temperature for
\hbox{V-4Cr-4Ti} alloys.

In conclusion, no irradiation effects have been observed which would limit
operation of \hbox{V-4Cr-4Ti} to $<100$~dpa as long as the component is
designed in accordance with rules for unirradiated material and the minimum
operating temperature is set at 430\SdegC.  Operation to
200~dpa may require a 10\% reduction in the allowable primary membrane stress.
However, the data base for \hbox{V-4Cr-4Ti} needs to be expanded to higher
neutron damage levels, higher helium levels and lower temperatures to support
these conclusions.  Also, fatigue data should be generated to allow a fatigue
design analysis to be performed.  The final issues which need to be addressed
are compatibility with the coolant and/or coating and the behavior of hydrogen
isotopes which may be implanted in, generated by transmutation in, or
transported to, the \hbox{V-4Cr-4Ti} structural material.

\subsubsection{Compatibility issues}
\label{s-6-3-3-4}

Compatibility of V-alloys with gaseous environments is important during
fabrication, handling and maintenance, as well as operation if gas ({\it e.g.},
He) is used as a coolant.  Compatibility with liquid Li is also important.  The
V-alloy/Li case is complicated by the presence of an electrically insulating
layer ({\it e.g.,} AlN or CaO) which is required for MHD pressure drop
reduction.  The behavior of implanted, permeated or transmuted hydrogen
isotopes must also be addressed.

The specifications for the reference \hbox{V-4Cr-4Ti} material call for
$<30$~appm H and $<1,000$~wppm of O, C, and N.  Significant increases in these
non-metallic impurities can cause undesirable decreases in the tensile
ductility and the impact DBTT.  The main problem is caused by high interstitial
levels of these impurities.  The solubility limits for O, N, and C impurities
in V-alloys at 500\SdegC are \hbox{$1.1\X 10^4$~wppm} for O, \hbox{$3.9\X
10^3$~wppm} for N and 400~wppm for C.  Hydrogen should be treated separately
from these other impurities.  At 1~Pa of H$_2$ pressure, the solubility of H in
V-alloys is about 400~appm (about 8~wppm).  For a gas-cooled system ({\it
e.g.,} He) impurity control, particularly for O$_2$ and H$_2$O has to be
exceedingly tight in order to prevent embrittling oxidation.  For a V-alloy/Li
system, the problem is much less severe because Li has a higher chemical
affinity for H and O than does V.  For example, in the temperature range of
280--550\SdegC, the ratio of H in the Li to H in the V-alloy is about 1000/1 if
the system achieves thermodynamic equilibrium.  Under these same conditions,
the ratio of O in the Li to O in the V-alloy is greater than 10 million to one.
Thus, impurity control for the V-alloy/Li system centers around controlling the
N and C levels in the coolant.  In particular, N levels should be limited to
$<20$~wppm and C levels should be limited to about \hbox{$6\X 10^{-8}$}~wppm in
Li.  There is no conceptual or practical problem in doing so.  These limits
would maintain the N level in the V-alloy to below 1800 wppm and would keep the
C level to below 200~wppm in the V-alloy.  The role which the MHD coating plays
in the impurity control and distribution of impurities needs to be addressed in
future R\&D work.

Impurity levels in the Li coolant can lead to enhanced corrosion of V-alloys, 
as well as to direct embrittlement of the V-alloy through transport of the 
impurities to the V-alloy.  In general, if the impurity levels are controlled 
such that C and N transfer to the V-alloy is within design limits, the 
corrosion rates are very low.  For a bare V-alloy/Li interface, the corrosion 
rate is only about 1 mm/year at 700\SdegC.  Thus, protecting the V-alloy from 
embrittlement eliminates corrosion as a concern.

Excessive tritium build-up in the V-alloy is of concern from both safety and
embrittlement perspectives.  These problems have been addressed in past design
studies.  However, there is a high degree of uncertainty associated with 
modeling the behavior of tritium implanted or transmuted in coated V-alloys.
Thermodynamically, the Li coolant has a much higher affinity for the hydrogen
isotopes than does V.  The central issue for coated V-alloys becomes the degree
to which the electrical insulator coating acts as a barrier to hydrogen-isotope
transport.  More work, both analytical and experimental, needs to be performed
in this area.

While most of the past R\&D has focused on V-alloy/Li compatibility, future
efforts need to be directed at the compatibility of the electrical-insulator
coating with Li, the compatibility of the coating with the V-alloy and the
compatibility of the overall system.

\subsubsection{Fabrication}
\label{s-6-3-3-5}

Vanadium alloys have been fabricated into a variety of shapes and sizes,
including plate, sheet, tubing and wire.  The major issue is to avoid
environmental contamination during processing.  Oxygen is a particular problem
as V-alloys react rapidly with oxygen at elevated temperatures~\cite{Smi94a}.
Within the fusion R\&D program, most fabrication has been limited to 
producing small test samples.  Recently, a 3-m-long tube (48-mm ID and 60-mm
OD) of \hbox{V-4Cr-4Ti} has been fabricated with a CaO coating for the purpose
of testing in a magnetic field  with flowing Li.  Also, \hbox{V-4Cr-4Ti}
has been selected as a structural component for the DIII-D Radiative Divertor.
This project will provide valuable experience with a larger size heat
($\sim1,000$~ kg), fabrication of different product forms, welding of similar
and dissimilar materials, and non-destructive testing techniques.

Within the U.S.\ fusion R\&D program, the welding of V-alloys has received a
great deal of attention.  In particular, gas tungsten arc (GTA) or tungsten
inert gas (TIG), electron beam and laser welding techniques have been applied
and characterized.  The GTA method presents particular problems in that a high
purity atmosphere is required.  Electron beam welding has produced some
excellent results without the need for post-weld heat treatment (PWHT).  Laser
welding has produced some good results as well.  The need for PWHT has not yet
been established.  Experience with a particular lot of \hbox{V-5Cr-5Ti} (BL-63)
proved inconclusive in that both the base metal and the GTA-welded section
improved in ductility with a PWHT at 950\SdegC.

In summary, it is recognized that experience with vanadium-alloy fabrication
and joining is limited.  This is being addressed within the R\&D program.

\subsection{Ability to Meet Demo Requirements}
\label{s-6-3-4}

Several of the Demo requirements may act as discriminators between the 
different engineering design options.  Three of them are highlighted here as 
they relate to the use of vanadium alloy in a power plant.  These are:

\begin{enumerate}

\item
\underline{Cost and Performance Requirements}.  Attractive features of vanadium
include its higher temperature limit and improved thermophysical properties,
which allow operation at higher plant efficiency and higher power density.
However, the high material costs may offset these advantages.  Analysis shows
that a vanadium system which uses the alloy only where needed, and which adopts
a recycling strategy, can be economically competitive.

\item
\underline{Safety, Licensing and Environmental Requirements}.  Vanadium alloys
have the potential for superior safety and environmental characteristics.
Accident safety and waste management are examined to determine whether the
aggressive Demo requirements can be met.

\item
\underline{Reliability and Availability Requirements}.  There is no data on the
reliability of fusion components in the fusion environment.  However, analogous
data from other technologies suggest that achieving the availability goal will
require either significantly reduced failure rates and/or some form of in situ
repair mechanism or system redundancy.

\end{enumerate}

\subsubsection{Cost and performance requirements}
\label{s-6-3-3-6}

One of the top level requirements for the U.S.\ Demo power plant states that
the cost of electricity must be competitive.  In 1995 dollars, the goal is
80~mill/kWeh and the requirement is 90~mill/kWeh.  Sub-tier requirements are
then derived from this requirement and goal.  In the evaluation of advanced
structural materials, attention has been given to two key sub-tier
requirements: (1)~the total capital cost should be less than \$5B, and (2)~the
gross thermal efficiency should be higher than 46\%.  These requirements have
been evaluated for both the V/Li and He/Li/V blanket options.

The V-alloy Li self-cooled design was adapted from the \AII design, modified to
reduce the use of vanadium, and examined using the ARIES System Code.  For this
case, the first wall/blanket cost was \$69M and the shield cost was \$239M.
Together, the first wall, blanket, and shield cost is approximately 15\% of the
total direct cost of the power plant.  This figure is based on the unit cost of
the V-alloy structure at {\$300/kg, ferritic steel structure (\HT) at \$68/kg,
and tenelon filler at \$25/kg in constant 1992 dollars.  The cost of the
vanadium in the shield has been reduced substantially as compared with earlier
designs by using steel filler and mechanical support structure wherever
possible. (See Sec.~\ref{s-6-2-4-3})

Vanadium alloy is a new structural alloy.  To date, there have not been major
industrial applications to support its development.  The three basic costing
elements for structural material (cost for production of different product
forms, cost for fabrication, and cost for installation) are still to be
developed for V-alloy.  Therefore, there is no statistical information
available to project a reliable unit cost for first wall, blanket, and shield
components.  Only a rough estimate on the lower and upper bounds of the unit
cost can be made.

Based on the application of advanced structural materials for fusion
devices~\cite{Col95}, it is observed that the product form cost will amount to
about 30\% of the unit cost of the final component, including labor.  Taking
the 1995 cost of advanced Ni-based structural materials as a starting point,
the product form cost is about \$100/kg.  If this is assumed to be the same for
V-alloy, then the lower bound of V-alloy unit cost can be assumed to be about
\$300/kg.  At the other end of the spectrum, based on the recently placed
DIII-D contract for the \hbox{V-4Cr-4Ti} alloy material necessary for the upper
divertor in the Radiative Divertor Program (RDP), the finished V-alloy
component cost is about \$600/kg~\cite{Col95}.  For Demo, this unit cost can be
lowered because of increased experience beyond the year 2020.  But when
compared to the DIII-D RDP component, which is a relatively simple water-cooled
design, the unit cost may become higher because of the Demo design complexity.
Considering this trade-off, it may be reasonable to assume that the upper bound
for the Demo V-alloy unit cost is about \$600/kg.

\subsubsection{Safety, licensing and environmental requirements}
\label{s-6-3-3-7}

Vanadium alloys have the potential for superior safety and environmental 
characteristics when compared with other near-term structural materials.  A 
safety evaluation of any structural material must include the impact of 
routine emissions, dose rates in working areas, decay heat, potential off-site 
doses from hypothetical accidents, and the management of end-of service 
material.  The two safety aspects of primary importance in assessing a 
material for use in the DEMO reactor are the consequences of accidents and the 
options for waste management.

\paragraph{Consequences of accidental releases.}

Potential doses from an accidental release of activated material depend on the
radioactive inventory, the volatility of the compounds containing the
radionuclides, and available means of transport for the radionuclides to the
facility boundary.  The isotopes governing early off-site doses are $^{48}$Sc,
$^{42}$Ar, $^{24}$Na, $^{46}$Sc, $^{45}$Ca and $^{51}$Cr.  The potential for
mobilization of specific elements from fusion-relevant materials in
high-temperature excursions under oxidizing conditions has been investigated in
experiments at INEL~\cite{But94}. In the vanadium alloy volatility tests,
vanadium alloys were doped with scandium, calcium and manganese to simulate the
activation products.  Based on the release rates observed in these experiments,
elemental mobilization fractions for the alloys \hbox{V-15Cr-5Ti} and
\hbox{V-3Ti-1Si} corresponding to a material thickness of 5~mm and duration of
1~h at elevated temperature were calculated.

In estimating the potential off-site radiation dose, the following pessimistic
conditions are assumed: (1)~2-km site boundary radius; (2)~1~m/s wind velocity;
(3)~Pasquill F stability conditions; (4)~near-ground-level release (0--10~m);
(5)~no release delay; (6)~1~h release duration; (7)~200~m height of mixing
layer; (8)~no rain; (9)~rough (urban or forest terrain roughness index); and
(10)~building wake effects for a $40\X 40$~m building.

Based on these conditions, the early doses to an individual at 2~km for designs
employing \hbox{V-15Cr-5Ti} and \hbox{V-3Ti-1Si} were about a factor 30 below
those for 316L stainless steel for a range of first wall temperatures from
900~K to 1,500~K.  With the \hbox{V-15Cr-5Ti} alloy, $^{42}$Ar dominates the
dose at the lower temperatures while $^{48}$Sc dominates at higher
temperatures.  It is assumed that all the argon is released over the 7-day
duration of the accident.  Other isotopes contributing more than 5\% to the
early dose are $^{46}$Sc, $^{45}$Ca, and $^{51}$Cr.

With the \hbox{V-3Ti-1Si} alloy, off-site doses are governed by $^{24}$Na at
lower temperatures and $^{48}$Sc at higher temperatures.  A conservative value
of 0.3, based on the properties of sodium at typical accident temperatures, was
used in the absence of experimental data on the mobilization fraction for
sodium.  Additional isotopes contributing more that 5\% to the early dose are
$^{46}$Sc, $^{45}$Ca, and $^{42}$Ar.

{\bf Options for Waste Management} -- The waste management options for vanadium
components are either recycling or direct disposal.  Over the period relevant
to waste disposal, the radioactive inventory of the vanadium alloys is
dominated by the argon isotopes $^{39}$Ar ($T_{1/2}=269$~y) and $^{42}$Ar
($T_{1/2}=33$~y), generated almost entirely from the titanium.  The argon
inventories at shutdown for the alloy \hbox{V-15Cr-5Ti} exposed in first wall
are \hbox{$4.9\X 10^{-7}$~g} of $^{39}$Ar/kg(V) [\hbox{$6.2\X 10^5$~Bq/kg(V)}]
and \hbox{$2.0\X 10^{-6}$~g} of $^{42}$Ar/kg(V) [\hbox{$1.9\X 10^7$~Bq/kg(V)}].
For the blanket, the inventories are \hbox{$2.2\X 10^{-9}$~g} of
$^{39}$Ar/kg(V) [\hbox{$2.7\X 10^3$~Bq/kg(V)}] and \hbox{$1.1\X 10^{-8}$~g} of
$^{42}$Ar/kg(V) [\hbox{$1.1\X 10^5$~Bq/kg(V)}].  Inventories for
\hbox{V-3Ti-1Si} are approximately 60\% of these.

Two possible mechanisms can be envisaged for release of the argon after
disposal: (1)~argon may diffuse out of the metal; and (2)~argon could be
released during corrosion.  The first mechanism is unlikely to dominate since,
if the argon has remained immobile in the alloy throughout 
high-temperature service, it is improbable
that the gas will escape in a repository where the temperature is not likely to
exceed 350~K. In assessing the second mechanism, relevant corrosion information
for the alloys is needed.  Since that information is not available for alloys,
the behavior of pure vanadium has been taken as a guide.  Although vanadium
should theoretically corrode under reducing, high pH conditions, in practice it
is found that its corrosion behavior is similar to that of a noble metal.  Thus
the direct disposal of vanadium alloy is tantamount to the employment of large
volumes of expensive, essentially inactive metal as a packaging medium for
extremely low concentrations of activation products.

The relative volatility of the main activation products suggests a recycling
process based on high-vacuum remelting or refining, with immobilization of the
liberated argon isotopes in a stable metal matrix of low volume for geological
disposal.  With adequate initial purity control, several usage cycles should be
possible, thus greatly reducing the consumption of vanadium alloy and the
volume of waste.  However, this approach requires geological isolation.
Moreover, the use of recycled material is found not to affect the low off-site
accident dose calculated for these vanadium alloys.

\subsection{Key Issues and R\&D Needs}
\label{s-6-3-5}

Three critical issues associated with the use of vanadium have been examined.
Since vanadium is not a common engineering material, its basic properties,
especially under 14-MeV neutron irradiation, are uncertain.  An extensive
review of the state of our knowledge of vanadium properties is provided in
\hbox{Sec.\ref{s-6-3-3}} and in the appendix.  A few key areas are highlighted
here.  Second, some form of electrical isolation between the coolant and
conducting walls is probably necessary for using of liquid metal as coolant in
a tokamak.  For the reference self-cooled design, a CaO coating is postulated,
with AlN as a backup material.  Research on MHD coatings is still in a very
preliminary stage, but this issue is critical for the success of this class of
concepts.  Third, tritium recovery and control from both Li and He have been
examined.  Recovery methods have been developed, but the stringent requirements
for impurity control (to prevent degradation of the vanadium properties) leads
to special concerns.

\subsubsection{Vanadium material R\&D}
\label{s-6-3-5-1}

Considerably more experience is needed in fabricating and joining of V-alloys.
The largest heat of \hbox{V-4Cr-4Ti} produced to date within the U.S.\ Fusion
Advanced Materials Program is 500~kg.  The anticipated heat for the DIII-D
radiative divertor project will be about twice that size.  The project offers
the potential for expansion of experience in fabricating different shapes and
welding of \hbox{V-4Cr-4Ti} to itself as well as other metals.  More work needs
to be done in characterizing and optimizing welding techniques, particularly
electron beam and laser welding techniques.  The degree to which a post-weld
heat treatment is needed must be determined because it has important design
consequences for both assembly and maintenance.

The larger heats (500 and 1,000~kg) of \hbox{V-4Cr-4Ti} offer the important
opportunity of generating a more statistically valid set of mechanical
properties data for the material in the unirradiated condition.  From a design
perspective, at least 20 data points should be generated at room temperature
for ultimate tensile strength, yield strength, uniform elongation and reduction
in area.  Several data points at 50\SdegC temperature intervals up to 700\SdegC
should be generated for these properties in order to establish temperature
trend curves.  Concurrent with the generation of baseline data for the existing
reference \hbox{V-4Cr-4Ti}, additional characterization of the tensile
performance needs to be done with respect to Si alloying additions,
non-metallic impurities (H, O, C, N) and heat treatment in order to optimize
the material from a design perspective.  In particular, a substantial
improvement would be realized if the ultimate tensile strength could be
improved while maintaining the DBTT at a value which is comfortably below room
temperature, maintaining the uniform elongation at $>5\%$, and the fracture
toughness and fatigue properties (both yet to be determined) at acceptable
levels.

In addition to baseline fracture toughness and fatigue data, thermal creep and
time-to-rupture data for \hbox{V-4Cr-4Ti} needs to be generated for
temperatures in the range of 600--700\SdegC and stresses in the range of
125--375~MPa.  The issue which should be addressed is whether or not thermal
creep limits the design allowable stresses during a period of 2--4~full-power
years.  Higher-stress/lower-time tests may have to be employed for practical
reasons.

The attractiveness of a Li/V design depends greatly on the success achieved in
developing a self-healing electrical-insulator coating.  In addition to the
materials tests which have already been performed, testing in a magnetic field
is essential.  Concurrent with this testing, laboratory-scale tests need to be
performed on non-metallic impurity ({\it e.g.,} O, C, N) distributions in a
system with coated \hbox{V-4Cr-4Ti} and flowing Li.  The remaining issue to be
addressed is the behavior of hydrogen isotopes in the same system.  Depending
on whether or not an additional plasma-facing material is used, deuterium and
tritium will be implanted into the V-alloy and protium will be generated within
the first-wall V-alloy at a rate as high as 62.5-at.\%/100-dpa.
Calculational models use parameters which lead to at least an order of
magnitude of uncertainty in predicting inventory and permeation rates.  More
data are needed in order for the designer to assess the safety issues related
to tritium retention and the performance issues related to hydrogen
embrittlement.

The design goal of 100 to 200~dpa for V-alloys requires considerable expansion
of the data base.  Data on other V-Cr-Ti compositions may be used to justify a
lifetime of about 100~dpa for first-wall and/or high-heat-flux components.
Shielding lifetimes may be considerably longer than first-wall lifetimes
because of lower He/dpa ratios, lower H generation and lower thermal stresses.
For the reference \hbox{V-4Cr-4Ti}, the published data base is for $\leq31$~dpa
and $\leq75$~appm He in the temperature range of 420--600\SdegC.  Higher
fluence data are required (at least up to 100~dpa) to see if the swelling
saturates at $\leq5$~vol.\%, to determine the irradiation creep rate, and to
insure adequate tensile and impact ductility.  Given that austenitic and
ferritic steels suffer substantial ductility losses when irradiated at
200--400\SdegC in a thermal or mixed neutron-energy spectrum designed to
generate 1--10~appm He/dpa, it is desirable to demonstrate that
\hbox{V-4Cr-4Ti} maintains its good uniform elongation and DBTT for
irradiation/test temperatures in the range of 300--400\SdegC.

The U.S.\ Advanced Materials Program has completed irradiation of
\hbox{V-5Cr-5Ti} specimens in EBR-II (COBRA-1A2) up to about 30 dpa in the
temperature range of 380--410\SdegC.  The industrial heat of \hbox{V-4Cr-4Ti}
has been irradiated to a low damage level (about 4 dpa) in EBR-II (X520) at
temperatures in the range of 375--400\SdegC.  Additional tests to higher damage
levels and/or lower temperatures are planned for the U.S.\ ATR facility and the
RF BOR-60 reactor.  The BOR-60 irradiation are scheduled for reactor-insertion
in August 1995.  They are estimated to be in the temperature range of
360--400\SdegC.  \hbox{V-4Cr-4Ti} tubes are being fabricated for insertion into
ATR in the temperature range of 200--300\SdegC to determine in-reactor creep
rates.  Longer-range ATR tests are in the planning stage for irradiation at
350, 500 and 650\SdegC.  Other facilities are also being considered.  The
difficulty in planning a meaningful test has to do with transmutations.
Thermal neutrons need to be filtered out of the spectrum to prevent excessive
Cr build-up.  The transmutations which generate He at levels as high as 6.5
appm/dpa in a fusion spectrum are difficult to simulate for the fission neutron
energies available.  Some success has been achieved in this latter area by
allowing tritium to diffuse from the lithium thermal bond within TZM capsules
into the V-alloy and then decay to $^3$He.  Aside from being facilities-limited
and possibly resource-limited, irradiation testing to design-relevant damage
levels is very time-consuming.

\subsubsection{Insulator coating development}
\label{s-6-3-5-2}

Recent analyses have demonstrated that self-cooled liquid metal blankets in
high-field tokamaks are feasible only with electrically insulating coatings at
the walls~\cite{Mal95}. Without coatings, the MHD pressure drop would be too
large to allow adequate heat removal.  Different coating materials have been
examined for both steel and vanadium structures.  For the steel-PbLi system,
Al$_2$O$_3$ is the preferred candidate, whereas for the V-Li system, both AlN
and CaO are considered attractive candidates.

Given that compatible coating materials have been identified, the principal
concerns with any coating include: (1)~fabrication technology; (2)~ensuring
adequate resistance and adhesion as-fabricated; (3)~ensuring adequate
resistance and adhesion under operating conditions (including irradiation and
stress fields); (4)~effects of flaws on thermal hydraulic performance; and
(5)~self-healing.

Self-healing is an important consideration because flaws which may develop
during operation could seriously degrade blanket performance and lead to
failures.  Material systems must naturally reestablish the coating by their
inherent chemistry.

At this time, CaO is the first choice in the U.S.\ liquid breeder blanket
development program as a coating material because it has higher electrical
resistivity than AlN, is more amenable to application in complex shapes, and
appears to offer superior rehealing characteristics.  Recent development work
has concentrated on improving the fabrication methods along with the properties
of CaO.

The fabrication of the CaO coating takes place in three steps: (1)~Oxygen
charging of the vanadium alloy surface (nominally, 650\degC in an Ar
atmosphere containing 10--20 ppm oxygen. This results in a layer 30--50~mm
thick that contains $\sim1$~at.\% oxygen.); (2)~Production of a pure Ca layer
on the vanadium surface (nominally, 500\degC in a Li-2at.\% Ca mixture.  This
results in a Ca coating 5-10~mm thick); and 3)~oxidation of the Ca layer
(nominally, 625\degC in an Ar atmosphere containing 5000 ppm oxygen).

The resulting coating has been examined microstructurally as well as tested
in Li~\cite{mattas}.  Initial measurements are
encouraging.  Additional small scale tests are now in progress to study the
effects of changes from the nominal coating conditions on the properties of the
coating.  A significantly expanded research program is needed to fully resolve
this issue.

\subsubsection{Breeder and tritium control}
\label{s-6-3-5-3}

\paragraph{Tritium recovery and control for lithium.}

A unique characteristic of the tritium chemistry with the lithium system is the
very high solubility of tritium in lithium.  At 500\SdegC and with a tritium
concentration of 1 appm, the tritium partial pressure over lithium is only
\hbox{$1.5\X 10^{-10}$~torr}.  Thus, tritium control in a lithium system is
usually not an issue.  The issue is how to recover tritium from lithium if the
tritium concentration is 1~appm.

Many processes have been proposed to recover tritium from
lithium~\cite{Mor95}. The most developed concept is called molten salt recovery
process~\cite{Mar75}. The key step of this process is to mix the lithium with a
molten salt, during which the tritium will be transferred from lithium to the
salt.  The tritium from the salt can be recovered by an electrolysis process.
All the steps of this process have been demonstrated in a laboratory scale.
The feasibility of reducing tritium concentration in the lithium to $\sim1$~appm
was demonstrated. A key concern for this process is that salt will be dissolved
in the lithium during the salt/lithium mixing process.  This salt will be
carried back to the fusion core.  The impact of the salt on the lithium
corrosion characteristics and lithium chemistry is uncertain.  Thus, the
applicability of this process for a fusion reactor is in doubt.

During the ITER Conceptual Design Activity, a tritium recovery process based on
cold trapping was developed to recover tritium from lithium.  The cold trap
process was demonstrated to be able to recover tritium from lithium to the
saturation limit, as shown in \hbox{Fig.~\ref{f-6-3-12}}.  The key step of the
process developed for ITER was to add protium in the lithium. The protium
concentration was increased to about 1,000~appm.  Thus, at 200\SdegC, Li(H+T)
was super-saturated and would precipitate out.  This process was identical to
the cold trap process developed for the fast breeder program.  In that process,
Na(H+T) precipitate out in the cold trap and no isotope effect was observed.
The Li(T+H) can be separated from the lithium by gravitational force, which was
also demonstrated in the breeder program (called a meshless cold trap).  The
Li(H+T) will decompose in a drained cold trap to hydrogen gas at 600\SdegC.
The T and H will be separated by an isotope separation system based on
cryogenic distillation. Because the saturation limit at 200\SdegC is 440~appm,
which is far above the design goal of 1~appm, the cold trap process was
abandoned for tritium recovery from lithium in ITER.

\begin{figure}[t]
\centerline{\psfig{file=sti-6-3-12.ps,height=4.2truein}}
\caption[f-6-3-12]
{Cold trap process parameters.}
\label{f-6-3-12}
\end{figure}

\paragraph{Tritium control and recovery from He-cooled designs}

Tritium control and recovery from He is not a simple process when the tritium
partial pressure requirement is as low as required.  The tritium permeation
across the first wall from the plasma has been calculated to be of the order of
10~g/d including the effect of a diffusion barrier.  With a permeation rate of
10~g/d, the tritium partial pressure increase is $10^{-2}$~Pa per pass.  If a
closed cycle gas turbine is used for energy conversion, a heat exchanger will
be required for heat rejection.  For a heat-exchanger temperature of 300\SdegC
and a surface area of 20,000~m$^2$, which is made of 316~SS (a material with
rather poor permeability of hydrogen), the tritium partial pressure has to be
controlled to be lower than \hbox{$3\X 10^{-7}$~Pa} to limit the tritium
permeation rate in the heat exchanger to no more than 10~Ci/d.  If a tritium
diffusion barrier can be provided on the surface of the heat exchanger, the
tritium partial pressure has to be maintained at \hbox{$3\X 10^{-3}$~Pa},
which is still lower than the vacuum of the plasma. Thus, the increase in
tritium partial pressure during each pass of the He-coolant across the reactor
is higher than that acceptable by the heat exchanger, {\it i.e.,} the entire
reactor coolant has to be processed to remove the tritium permeated across the
first wall.  Since the flow rate of the He-coolant is much larger than that for
the purge gas, this process could be very expensive.

\begin{figure}[b]
\centerline{\psfig{file=sti-6-3-13.ps,height=4.0truein}}
\caption[f-6-3-13]
{Tritium process system for a solid breeder blanket.}
\label{f-6-3-13}
\end{figure}

For He-cooled systems with liquid lithium breeder, tritium recovery from the
breeder is similar to that of self-cooled Li/V designs.  For systems with solid
breeders, a purge system has to be utilized.  Figure~\ref{f-6-3-13} shows the
tritium process system for a solid breeder blanket.  The purge stream will have
to be cooled to $\sim100$\SdegC to cold trap the hydrogen isotopes in the
purge.  Due to the small mass flow rate of the purge gas in this case, the
power requirement and the size of the equipment for this step is acceptable.
However, this step could be excessively expensive if the entire coolant has to
be processed this way. Tritium also can be recovered by an oxidation process to
form HTO.  The cold trap temperature to recover HTO does not have to be as low
as for the HT.  However, this method will introduce excess oxygen and cannot be
recommended for the V-system.

\section{SILICON-CARBIDE COMPOSITE DESIGNS}
\label{s-6-4}

\subsection{Design Options}
\label{s-6-4-1}

Silicon-carbide (SiC) composite material with a projected allowable temperature
capability $\geq$ 1000\SdegC was selected as the structural materials for
\AI~\cite{ARIES1} and \AIV~\cite{ARIES4} reference and alternate blanket and
divertor designs.  These designs have been used as the basis for assessing the
advantages and issues associated with a possible Demo power plant based on SiC
composites as the primary in-vessel structural material.

To take full advantages of the low activation and high temperature capability
of SiC composite material, 5--10~MPa helium was used as the coolant.  \AI used
Li$_2$ZrO$_3$ as the solid tritium breeder, Be metal sphere-pac as the neutron
multiplier, and W as the divertor coating material.  Based on new experimental
results, Li$_2$O was selected as the solid breeder for the \AIV designs (to
eliminate Zr activation issue).  If the use of W can be avoided as the divertor
coating material, then an inherently safe design (LSA~=~1) can be
obtained.  This also assumes that the Be chemical energy and toxicity concerns
are resolved.

Both \AI and \AIV alternate designs used the toroidal flow, nested 
shell blanket configuration, where the SiC composite components can be made 
from smaller ceramic (about 1~m~$\X$~1~m) parts.  Reliability of this 
configuration depends on the successful development of ceramics joining 
techniques.  The \AIV reference design has a poloidal flow configuration.  
This relies on the successful development of the fabrication of large 
($>1$~m wide, $\sim1$~m deep and $>7$~m in height components) leak-tight 
SiC composite components.

Both \AIV reference and alternate blanket designs satisfy all of the neutronics
and thermal hydraulic performance requirements of Demo.  Except for the need of
Be-neutron multiplier, both designs will have excellent safety characteristics.
We also found that by increasing the helium pressure to about 12~MPa, and
taking advantages of the improved recuperator performance of the closed cycle
gas turbine system, at a helium outlet temperature of about 950\SdegC, a gross
thermal efficiency of about 55\% can be expected.  In the evaluation of the use
of SiC-composite as the structural material for the Demo design, there was no
doubt of its projected benefits; the key question is on whether its development
schedule can meet the projected schedule of Demo given the small R\&D effort on
this material in U.S.

\subsection{Key Issues}

Key issues for the application of SiC composite are in the areas of material
development that can match the schedule of the U.S.\ Demo design, the behavior
of the composites in a fusion environment, the need for electrically-conducting
components for plasma stability, and the development of robust plasma facing
components (PFC). Fundamental improvements in material irradiated properties,
such as thermal conductivity, are needed, for example by using advanced SiC
fiber and interface materials.  Other material development issues include
economic fabrication of large SiC-composite components, development of vacuum
leak~tight components, techniques for brazing ceramic parts, and the
development of the joining techniques of SiC~composite to metallic parts.

Most advanced tokamak scenarios require a close-fitting conducting shell to
stabilize kink modes.  A separate shell may be needed behind the blanket for
passive stabilization of axisymmetric modes (vertical stability). If metallic
shells should be used in regions of high-neutron fluence, the safety rating and
thermal performance of the plant may be impacted.  

Various techniques of leveling the surface heat flux, such as the use of the
radiative divertor approach is essential, since it dictates the feasibility of
removing surface heat flux by helium coolant and the utilization of low
activation material for the PFC surface and components.  The concurrent
reduction in the ion energy to $<5$~eV is necessary to maintain low surface
erosion and therefore, adequate component lifetime and to maintain the low
activation benefits of the SiC composite design without the use of high-$Z$
surface material, such as W.

\section{MAGNET SYSTEMS}
\label{s-6-5}

Several magnet issues related to licensing and commercial acceptance of fusion
tokamak power plants have been addressed. First, several types of faults on
both the poloidal and the toroidal field systems were analyzed.  Results of
these studies are presented in Secs.~\ref{s-6-5-1} and \ref{s-6-5-2}. Second,
because the maintainability and availability are important Demo requirements,
access to the blanket and shield as well as removal and repair of the magnet
system have been considered.  Section~\ref{s-6-5-3} covers discussion of
several methods to provide the dewar/cryostat system for the toroidal field
system, with the investigation addressing the relevance of the possible designs
to the maintenance of the magnet system and the blanket/shield.  Finally,
Sec.~\ref{s-6-5-4} deals with the structural issues on the toroidal field coil
that arise from enlarging substantially the access port in the outer leg of the
toroidal field coils.

\subsection{PF Fault Analysis}
\label{s-6-5-1}

In this section some failure modes of the poloidal-field coils are analyzed.
Since detailed parameters necessary for this type of calculation were not
available, a well documented design, TPX was chosen for carrying out the
evaluation.  Parameters not in that report were obtained from scientists at
Princeton Plasma Physics Laboratory.  One of the reasons for analyzing TPX
rather than ITER was that the Demo, as well as TPX, are up-down symmetric
(up-disruptions are different in ITER from down-disruptions).

In order to determine the worst loads, it is necessary to carry out an
evaluation for a full, time-dependent discharge scenario.  Two classes of
failure modes were analyzed, each starting with a plasma disruption.  In the
first class, a PF system dump occurs some time after the disruption.  In the
second class, a short circuit in the terminal of only one PF coil occurs some
time after the disruption.  This short circuit is then followed by discharge of
the remaining PF coils through their dump resistors.  Obviously the number of
combinations of these fault scenarios is very large.  For the first class of
faults, the PF coil dump can occur at any time after the disruption (three
delay times of 0, 0.5, and 1~s were considered here).  For the second class of
faults, the short can occur in any PF coil (short circuit of all 14 PF coils,
one at a time, were analyzed) and at any time after the disruption (three delay
times of 0, 0.5, and 1~s were considered here).

The magnet system should be designed so that it can handle the largest load in
a given coil or element occurring during normal operation or faults.  Therefore,
for each fault discharge scenario outlined above it is necessary to calculate
the stresses in all the coils and then to determine the coil with the highest
stress and/or force. Due to the large number of items (stresses, forces) that
need to be monitored, a limited set of parameters were chosen to represent the
complex time-dependent forces and stresses in the coils.  They are: (1)~Maximum
compressive forces on the central solenoid (CS) interfaces; (2)~Maximum
tensile (separating) forces on the CS interfaces; (3)~Maximum differential hoop
force per turn between the CS coils; (4)~Maximum differential hoop force per
turn between the end coils of the CS and the adjoining retaining plates;
(5)~Maximum compressive hoop force per turn in any individual coil of the CS;
(6)~Maximum tensile hoop force per turn in any individual coil of the CS;
(7)~Maximum and minimum vertical body forces on the CS stack; and (8)~Maximum
tensile/compressive/vertical/centering loads in each of the poloidal field ring
coils (excluding the coils in the CS).  In addition, the effects of PF coil
failure on the toroidal field (TF) coils was estimated by monitoring the
following: (1)~Maximum out-of-plane force on either upper or lower TF coil;
(2)~Maximum out-of-plane moment on either upper or lower TF coil, calculated
about the equator of the machine; and (3)~Maximum out-of-plane moment on either
upper or lower TF coil, calculated about the axis of the machine. In total, 23
quantities were calculated and monitored.

The axisymmetric model used in the calculations included the plasma, the PF
coils, the inner coils, and the passive structure.  The TF coils are modeled
with 3-D lumped current filaments carrying a current of 2.8~MA. The programs
EIGENCIRC~\cite{Rad95} and SOLDESIGN~\cite{SOLDESIGN} with postprocessors for
force calculations were used in the studies.  It was assumed that some of the
coils are connected in series.  The plasma disruptions were modeled using 9
discrete plasma positions.  It was assumed that the time constant for the
plasma drift prior to the disappearance of the plasma current (disruption) was
0.1965~s.  It also was assumed that the plasma current disappeared
instantaneously at the end of the plasma drift.  The PF coils were flux
conserving during the disruption.  The current in the passive elements were
assumed to be zero prior to the disruption.  In order to include the finite PF
currents due to currents in the passive elements after the plasma current
decays to zero, the calculations were carried out 5~s after the end of the
plasma decay.  By this time the currents in the passive elements were
essentially zero.

\begin{table}[p]
\caption{}
\centering{\bf Maximum Normalized Force or Stress During the PF Fault\\
(Normalized to the Maximum Normal-Operating Quantity)}
\vspace{1.cm}
\renewcommand{\arraystretch}{1.05}
\begin{tabular}{l @{\hspace{1cm}} c}
\hline\\
Central Solenoid: \\
\quad Compressive                         & 1.03  \\
\quad Separation                          & 1.24  \\
\quad Diff. hoop/turn on coil interface   & 2.97  \\
\quad Diff. hoop/turn on retaining plates & 1.81  \\
\quad Hoop/turn - compressive             & 2.40  \\
\quad Hoop/turn - tensile                 & 2.13  \\
\quad Vertical body - upward              & 7.63  \\
\quad Vertical body - downward            & 1.23  \\
PF5: \\
\quad Hoop - compressive   & 12.12 \\
\quad Hoop - tensile       & 1.10  \\
\quad Vertical - centering & 1.00  \\
\quad Vertical - launching & 1.24  \\
PF6: \\
\quad Hoop - compressive   & 1.34  \\
\quad Hoop - tensile       & 1.00  \\
\quad Vertical - centering & 1.17  \\
\quad Vertical - launching & 1.76  \\
PF7: \\
\quad Hoop - compressive   & 1.70  \\
\quad Hoop - tensile       & 1.01  \\
\quad Vertical - centering & 1.37  \\
\quad Vertical - launching & 1.00  \\
TF: \\
\quad Side force on half of TF coil      & 1.23  \\
\quad Moment on TF coil about equator    & 1.37  \\
\quad Moment on TF coil about major axis & 1.71  \\
\\ \hline
\end{tabular}
\label{t-6-5-1}
\end{table}

In order to summarize the results in the most compact and convenient manner,
the maximum stress or force was normalized to the maximum loading during normal
operation. This would indicate safety margins that should be included in the
coil designed for normal operating conditions. \hbox{Table~\ref{t-6-5-1}} shows
the worst-case values for the 23 monitored quantities over all the fault
discharge scenarios. The results indicates that with a few exceptions, a design
based on the normal scenario operation is robust enough to handle the fault
conditions.  These exceptions include: (1)~Hoop loads in the central solenoid;
(2)~Launching force in the central solenoid; and (3)~Launching force on the
divertor coil (PF5). The support of the compressive loads can be provided, and
should be of interest mainly for preventing buckling of the coils.

\subsection{TF Fault Analysis}
\label{s-6-5-2}

The fault analysis of the TF coils was carried out in an ITER-like system,
since the toroidal field coils are more similar to those presently being
considered for the DEMO with poloidal shells, as opposed to radial plates (as
in the \AI design) or with a conductor winding inside of a case (as in TPX).
The fault analyzed here is one in which a short is postulated across the
terminals of one of the TF coils and the rest of the TF coils discharged
through their dump resistors.  This is a plausible failure mode as it could
happen either when a coil short is detected and the TF coil system is shutdown,
or when the short occurs shortly after the beginning of a TF coil dump. 

Circuit simulations of a concurrent dump of the remaining coils results in a
very high overcurrent in the shorted coil.  A quench develops before the full
development of this current. The coupling between the circuit equations and the
thermal hydraulic calculations that model the quench behavior of the conductor
is complicated.  If the shorted coil does not quench, the current in the
shorted coil increases to 350\% of its nominal value during the discharge of
the rest of the TF system.  However, with these currents at TF field, the coil
is likely to quench.  It was determined that 5~s into the coil dump, the field
and current in the shorted coil were such that the conductor went into the
current sharing mode (with current in the normal conductor in parallel with the
superconductor) at the operating temperature.  A quench was postulated at this
time, and the circuit calculation carried out from this point with ``internal''
dump of the shorted coil.  It was assumed that either 10\% or 100\% of the
shorted coil was in the normal state, or that 300~m of the conductor quenched
with quench propagation.  These assumptions will have a greater effect on the
ultimate temperature of the conductor rather than on the worst case structural
response. For the cases analyzed, the pack current did not increase beyond
100~kA (about 250\% nominal value).

Understanding the consequences of the entire history of this event will require
further simulation, and finding the limiting structural condition will require
the analysis of the time of the PF scenario and the decision to discharge the
PF coil.  In order to estimate the structural response, two points in this
scenario are analyzed.  Point~(A) is early in the fault when the PF coils are
at their full end-of-burn currents. It was assumed that the current in the
shorted toroidal field coil was at 120\% of nominal current while the currents
in the other TF coils were decreased to 80\% of the nominal value.  Point~(B)
is later in the fault, when the PF coils are de-energized.  At this point,
there is a large overcurrent in the shorted coil and the remaining TF coils
also have substantial currents, which produce significant structural
interaction with the shorted coil.  For this case, it was assumed that the
current in the shorted TF coil was at +250\% of the nominal current while the
currents in the rest of the TF coils were at 50\% of nominal value.

A simplified model of the entire tokamak has been constructed.  The TF coil has
been constructed with beam elements, and coarser plate and solid elements have
been used throughout.  The modeling approach eliminates credit for symmetry and
adds non-linear gap elements where major structural components interact.  The
model is intended to be simpler than models used previously for NET, but
sophisticated enough to allow the major phenomena to be identified for a full
complement of loading and structural interactions.  The keys are modeled as a
continuous shell of plate elements.  This was intended to simulate the
torque-shell behavior of the assembly of keys and TF plates.  It also simulates
the toroidal bearing stress which results if neighboring coils are attracted to
one another.

\begin{figure}[p] 
\vbox to 3.2in{\vfil
\centerline{\psfig{file=sti-6-5-01.ps,width=5.25truein}}
\vfil}
\caption[f-6-5-01]
{Bending and tension for the four corners of the beam elements for all 24 coils
of the model at Point~(A) during the fault.}
\label{f-6-5-01}
\centerline{\psfig{file=sti-6-5-02.ps,height=3.5truein}}
\caption[f-6-5-02]
{Bending and tension for the four corners of the beam elements for all 24 coils
of the model at Point~(B) during the fault.}
\label{f-6-5-02}
\end{figure}

\begin{table}[t]
\caption{}
\centering{\bf Analysis of TF Coil Loads During Fault Conditions}
\vspace{1.cm}
\renewcommand{\arraystretch}{1.3}
\begin{tabular}{l c c}
\hline 
               &                & Upper-Half  \\
               & Centering (MN) & Bursting (MN)  \\
\hline \\
Nominal load \\
\quad Global Model      & --318 & 313  \\
\null\\
Point (A): \\
\quad Stick model       & --278 & 336  \\
Point (B): \\
\quad Single coil model & --411 & 700  \\
\quad Stick model       & --372 & 731  \\
\\ \hline
\end{tabular}
\label{t-6-5-2}
\end{table}

The loads in the toroidal field coils are shown in Figs.~\ref{f-6-5-01} and
\ref{f-6-5-02} and \hbox{Table~\ref{t-6-5-2}}.  The centering load refers to
the radially inward load in the vertical section of the coil (inboard side).
The upper half-bursting load refers to the integrated vertical load of the
upper half of the coil.  Obviously, point~(B) produces the worst condition with
a much increased vertical bursting loads. These loads result in increased
tension in the coil.  In addition, there are very large bending stresses in the
shorted coil (Figs.~\ref{f-6-5-02}).  This is basically due to the fact that
the system is far away from bending free, since the shorted coil is about twice
the current and the neighboring coils are half the nominal value.  The coil,
therefore, attempts to become bending free, especially since the poloidal
shells of the TF coils have very limited capability for carrying bending
stresses.  Only the intershell bond can carry much bending.  The large bending
stresses depend therefore on the integrity of the plate-to-plate bond and will
probably be relieved by bond failure. The shorted coil (with 250\% overcurrent)
has bending stresses of about 600~MPa, while the other coils are only slightly
loaded.

In summary, the failure modes analyzed here result in large deformations of the
shorted toroidal field coil.  It is, therefore, necessary to modify the design
in order to be able to support the bending loads that result from this failure
mode.  This issue is addressed in Sec.~\ref{s-6-5-4}.

\subsection{Cryostat and Dewar Design}
\label{s-6-5-3}

The maintenance scheme for \AI and \AIV designs included removal of an entire
fusion core sector (1/16 of the torus corresponding to 16 TF coils).  In \AI,
a separate dewar and cryostat was used for each coil.  The TF coil was removed
with a sector during the maintenance operation.  In \AIV, the entire TF coils
were located in one cryostat. The outer leg of the TF coils were extended to
allow removal of an entire sector radially in between the TF coils.  This
latter option has a larger and, thus, more expensive TF system. In this section
feasibility of separate dewar/cryostat for each coils are investigated.

\begin{figure}[p] 
\centerline{\psfig{file=sti-6-5-03.ps,height=3.1truein}}
\caption[f-6-5-03]
{Design with individual cryostat over each coil, with no thermal
insulation between adjacent cold coil elements (coil-to-coil, or coil-to-cap).}
\label{f-6-5-03}
\vspace{1.cm}
\centerline{\psfig{file=sti-6-5-04.ps,height=3.1truein}}
\caption[f-6-5-04]
{Design with individual cryostat over each coil, with thermal
insulation between adjacent cold coil elements (coil-to-coil, or 
coil-to-cap).}
\label{f-6-5-04}
\end{figure}

The \AI design utilized a partial cryogenic dewar as shown in
Fig.~\ref{f-6-5-03}.  There are insulating vacuum gaps only in the
regions that face the warm regions.  In regions that face other cold regions,
only the vacuum boundary is present without the presence of insulating material
(such as multi-laminar insulation, or superinsulation).  The gaps in-between
vacuum boundaries shown in Fig.~\ref{f-6-5-03} will fill up with
frozen material: ice and frozen oxygen or nitrogen.  It was expected that the
thermal conductivity of the slush that is formed in this manner would have very
low thermal conductivity.  In this manner it would be possible to prevent
warm-up of the regions of the coil that are adjacent to the corner region that
faces both a warm and cold region.

The alternative is shown in Fig.~\ref{f-6-5-04} with large insulating gaps all
around the coil. The main advantage of this approach is the absence of large
cross sections of the coil not filled with either structural material or
conductor.  In addition, it is possible to transfer loads effectively across
the coil boundary (from coil to coil, or from coil to bucking cylinder or
out-of-plane structure).  The superinsulation or G-10 spacers are good for
transferring the load in compression, but are not very effective for
transferring shears.

\begin{figure}[b] 
\centerline{\psfig{file=sti-6-5-05.ps,height=4.0truein}}
\caption[f-6-5-05]
{Phase diagram of ice.  Broken lines represent presumed phase boundaries which
have not yet been fully investigated.  Dotted lines represent metastable
continuations of one phase into a neighboring region.}
\label{f-6-5-05}
\end{figure}

Volume changes occur due to phase transformation and due to volume expansion as
water goes from liquid to solid (a 9\% increase in volume).  This expansion
would place both the magnet and the water in compression. As the temperature
decreases, there are phase transformations of the ice.  A phase diagram at 4~K
is shown in Fig.~\ref{f-6-5-05} as a function of the compressive
stress.  The phase Ih has a hexagonal crystal structure with density of 0.92
g/cm$^2$.  The phase IX (and III) has a tetragonal crystal structure with a
density of 1.14 g/cm$^2$ (as measured at -175C and 1 bar).  This means a 24\%
volume reduction if phase changes from Ih to IX at the same conditions.

Stress strain studies were carried out at MIT in order to investigate the
behavior of the ice.  Tests were performed at a MTS machine with a liquid
helium cryosystem.  The ice in the test chamber was made from local (Cambridge)
tap water by fast cooling with liquid nitrogen.  Partial results of
stress-strain tests are shown in \hbox{Fig.~\ref{f-6-5-06}}~\cite{Fen93}.  The
results at different load rates and at different temperatures show the same
features: (A)~the load is released repeatedly, starting from a critical stress
in the loading process; (B)~the release period increases as the load increases;
(C)~the released load value increases as the load increases; (D)~a similar
phenomenon with load jumps occurs in the unloading process.  The directly
measured strain data of ice as a function of stress has provided support that
these features are material properties of the ice.  The behavior of ice in
unloading process further indicates this features are due to the phase
transformation of ice.

\begin{figure}[b] 
\centerline{\psfig{file=sti-6-5-06.ps,height=3.8truein}}
\caption[f-6-5-06]
{Ice test results at a load rate of 10~lb/s and at 77~K}
\label{f-6-5-06}
\end{figure}

As a load is applied on a cylinder of ice, there is a stress gradient due to
the wall friction.  As the load increases to a critical number, a top part of
the ice first reaches the critical transition point, a phase change of Ih to IX
leads to a volume reduction of ice and to a load release.  As the load
increases, the portion of ice reaching at the critical point increases.  As the
load is decreased, the process is symmetrical.

\begin{figure}[b] 
\centerline{\psfig{file=sti-6-5-07.ps,height=4.2truein}}
\caption[f-6-5-07]
{Thermal conductivity of ice at cryogenic temperatures.}
\label{f-6-5-07}
\end{figure}

The thermal effects of the ice layer have been explored.  Figure~\ref{f-6-5-07}
shows the thermal conductivity of ice at the low temperatures~\cite{Sla80}.
The thermal conductivity of ice increases with decreasing temperature
(approximately inversely with temperature to about 10~K).  At cryogenic
temperatures it is about 3 orders of magnitude larger than the heat
conductivity of G-10, a good low temperature insulation material.  This implies
that the gap between adjacent coils has to be small.  It will be necessary to
add insulation at the side of the coils, as shown in
\hbox{Fig.~\ref{f-6-5-04}}.

A preliminary design of the insulation has been performed.  There are three low
thermal conductivity paths linked in series by two heat sinks stationed
nominally at 20 and 80~K.  The out-of-plane structure need not be symmetric
because the loads are asymmetric. The typical thermal conductivity integrals of
G-10 in the temperature range 4.2 to 300~K is calculated.  The values for the
three temperature ranges are shown in \hbox{Table~\ref{t-6-5-3}}.

\begin{table}[t]
\caption{}
\centering{\bf Thermal Conductivity Integrals and Heat Balance Conditions}
\vspace{1.cm}
\renewcommand{\arraystretch}{1.35}
\begin{tabular}{c  c c}
\hline 
Temperature range (K) && Thermal conductivity integral (W/m) \\
\cline{1-1} \cline{3-3}
\\
    \Z20 to \Z4.2     &&   \Z\Z2.212   \\
    \Z80 to  20\D     &&    \Z20.1\Z\Z \\
     300 to  80\D     &&     165.0\Z\Z  \\
\\ 
 Temperature (K)      && Net heat in (W/m$^2$) \\
\cline{1-1} \cline{3-3}
 20                   && \D357.8 \\
 80                   && 2,898\D \\
\\ \hline
\end{tabular}
\label{t-6-5-3}
\end{table}

The heat sink consists of stainless steel plates grooved and welded.  Helium at
30~atm is circulated as the coolant.  The G-10 spacers are on the order of 5~cm
thick.  The heat balance condition in the two heat sinks are shown in the
bottom part of \hbox{Table~\ref{t-6-5-3}}. In both these cases the cooling of
the heat sinks by helium at 30 atmospheres presents no problem.  Both the
temperature rise in the helium and the pumping power are small.

The above discussions show that (1)~Design with a cryostat as in
\hbox{Fig.~\ref{f-6-5-03}} is not attractive; and (2)~Design with a cryostat as
in \hbox{Fig.~\ref{f-6-5-04}} is feasible, but requires about 15~cm all around
each toroidal field coil. The 0.15-m dewar boundary around the each TF coil
requires a large fraction of the coil cross section of the TF coil in the
inboard side of the machine which increases the cost of the device
dramatically. Therefore, it appears that the most cost-effective method is to
locate the entire TF system in a cryostat and extend the outer leg of the TF
coils to allow removal of an entire sector radially in between the TF coils.
In the next section, the impact of these large access ports in between the TF
coils are discussed.

\subsection{Structural Effects of Large Access Port }
\label{s-6-5-4}

In this section a structural model is used to investigate the different methods
of supporting the Lorenz loads in the toroidal field magnet.  In particular,
the out-of-plane loads require special attention because the structures used to
support them interfere with the maintenance of the fusion core. A similar
approach to the one described in \hbox{Sec.~\ref{s-6-5-2}} is used in this
section.  Only normal conditions (end-of-burn) have been analyzed.  Four models
have been investigated, aiming at removing most material in-between the
outer legs of the toroidal field coils.  The outer leg of the toroidal field
coil has been moved sufficiently back so that an entire fusion core sector can
be removed through the gap in-between the coils with a purely radial motion.

The First model is shown in Fig.~\ref{f-6-5-08}.  It includes a cap (attached
to the bucking cylinder), but no additional structure in the region of the
outer legs of the TF coil.  The thickness of the cap is the same as the
thickness of the bucking cylinder.  The cap is attached to the toroidal field
coil through a key that runs along the toroidal field coil and the cap.  In
this manner, the TF coil and the cap are locked in the toroidal direction but
the coil is allowed to slide in the poloidal direction.

The stresses in the toroidal field coil are shown in Fig.~\ref{f-6-5-09}.  The
peak stresses (assuming smeared properties for the toroidal field coils) are
1.4~GPa in the area where the cap ceases to support the toroidal field coil.
In this case, the caps are rotating, taking along the coil (with
little capability to prevent the motion).  The TF coil, therefore, experiences
a strain-controlled condition, and the coil takes on the typical S-curve shape.
The coil experiences very large stresses across the region where the bending of
the coil is maximum.  In addition to the large loads in the region where the
cap ends and the TF coil is self-supporting, there are large in-plane bending
stresses at the top-bottom inner chord of the coil, as the coil and the
structure react to the large displacement of the caps.  At the inside surface,
top and bottom, the stresses are 1.2~GPa.  These stresses are very large, and
need to be reduced.

\begin{figure}[p] 
\centerline{\psfig{file=sti-6-5-08.ps,height=3.1truein}}
\caption[f-6-5-08]
{First model, with caps but no structure behind the outer leg of the TF coil.}
\label{f-6-5-08}
\vspace{1.cm}
\centerline{\psfig{file=sti-6-5-09.ps,height=3.1truein}}
\caption[f-6-5-09]
{Stresses in the TF coil for the first model, with 1.4~GPa maximum
stress in the region where the cap stops, and in the inner chord of the coil,
top and bottom, due to bending.}
\label{f-6-5-09}
\end{figure}

The second model, shown in Fig.~\ref{f-6-5-10} includes a reinforcement of the
TF coil, by placing structure directly behind the outer leg of the TF coil.
This structure is connected to both the outer leg of the coil and to the upper
and lower caps.  The goal is to increase the structural material in the outer
section of the TF coil in order to increase the moment of inertia, and to
decrease the rotation of the caps by reacting through shear stresses in the
structure directly behind the outer leg of the TF coil.  The thickness of the
structure in this region is the same as the thickness of the cap (which is the
same as the thickness of the bucking cylinder). In addition, the cap is attached
to the bucking cylinder both for rotation and for tension.  In this manner, the
cap helps support the in-plane tension of the coil, and provides for rigidity in
the case of TF-coil failure (short of a TF coil followed by dumping of the rest
of the TF coils, see Sec.~\ref{s-6-5-2}).

\begin{figure}[t] 
\centerline{\psfig{file=sti-6-5-10.ps,height=3.2truein}}
\caption
{Second model, with caps and structure (same thickness as the caps)
directly behind the outer leg of the TF coil.}
\label{f-6-5-10}
\end{figure}

\begin{figure}[p] 
\centerline{\psfig{file=sti-6-5-11.ps,height=3.2truein}}
\caption
{Stresses in the toroidal field coil for the second model, with 600~MPa bending
stress in the region near the discontinuity of the cap.}
\label{f-6-5-11}
\vspace{1.cm}
\centerline{\psfig{file=sti-6-5-12.ps,height=3.2truein}}
\caption
{Stresses in the external structure for the second model, with 2-GPa bending
stress in the structure directly behind the outer leg of the TF coil.}
\label{f-6-5-12}
\end{figure}

The coil stresses for the second model are shown in \hbox{Fig.~\ref{f-6-5-11}}.
The stresses in the region where the cap stops being toroidally continuous (and
where the stresses are 1.4~GPa in the previous model), are reduced to about
600~MPa.  In addition, the stresses in the cap near the top and bottom of the
machine are small, less than 60~MPa.  The cap thickness has been kept constant,
however, because making it thinner would result in larger rotation of the cap
and coil in the outer legs of the toroidal field coil.  It could be made
thinner, however, by using truss elements, or by placing structure in-between
toroidal field coils in the region underneath the caps.  The peak stresses in
the structure directly behind the outer leg of the toroidal field coil are
about 2~GPa, as shown in \hbox{Fig.~\ref{f-6-5-12}}, in the region close to the
location where the full cap ceases to be toroidal continuous.  It is necessary,
therefore, to further reinforce the structure directly behind the outer leg of
the coil, as described below.

In order to decrease the stresses, the structure directly behind the outer leg
has been further increased.  Two models of this case have been developed.  In
the first one (\hbox{Fig.~\ref{f-6-5-13}}), the height of the caps is such that
the inner components of the fusion core can be removed by simple radial
motions.  However, there is insufficient space for piping.  A fourth model,
shown in \hbox{Fig.~\ref{f-6-5-14}} was developed with even taller ports.  The
stresses in the external coil structure for the third model are shown in
\hbox{Fig.~\ref{f-6-5-15}}.  The stresses are acceptable with the exception of
the region in the inboard section of the structure in the outer leg, next to
the region of discontinuity of the caps.  The stresses in this region are on
the order of 1.2~GPa (down from 2~GPa).  This is due to the large displacement
of the caps.  If the port height were to be increased, the radius of curvature
needed to deform the coil and the structure directly behind the outer leg would
decrease, as would the stresses.  However, this would occur at the expense of
increased displacement (rotation).

Figure~\ref{f-6-5-16} shows the stresses in the TF coil for the fourth model
(with larger ports).  Again, the stresses peak in the region above and below
the coil, but the peak stresses are only 450~MPa.  The stresses in the external
coil structure (cap, bucking cylinder and coil-shadow) are shown in
\hbox{Fig.~\ref{f-6-5-17}}.  The stresses have decrease to about 700~MPa in the
structure.  However, increasing the height of the port (even though it
decreases the stresses), increases the displacements of the cap and toroidal
field coil.  The displacements are shown in \hbox{Fig.~\ref{f-6-5-18}}.  The
maximum displacement in this case is 0.05~m, while in the previous case (third
model) the maximum displacement was less than 0.04~m.

In conclusion, the alternatives for supporting the TF coils have been analyzed,
and one design with ports large enough for radial access to fusion core sectors
has been evaluated.  The structure needs to be poloidally and toroidal
continuous.  The caps will be keyed together, and attached to the bucking
cylinder with an arrangement similar to that in the ITER design (hole and pin).
The performance of this design should be established through further detailed
analysis.

\begin{figure}[p] 
\centerline{\psfig{file=sti-6-5-13.ps,height=3.2truein}}
\caption[f-6-5-13]
{Third model, with caps and reinforced structure directly behind the outer leg
of the TF coil.}
\label{f-6-5-13}
\vspace{1.cm}
\centerline{\psfig{file=sti-6-5-14.ps,height=3.2truein}}
\caption[f-6-5-14]
{Fourth model, with increased port height and reinforced structure directly
behind the outer leg of the TF coil.}
\label{f-6-5-14}
\end{figure}

\begin{figure}[p] 
\centerline{\psfig{file=sti-6-5-15.ps,height=3.2truein}}
\caption[f-6-5-15]
{Stresses in the external coil structure for the third model, with 1.2~GPa
maximum stress in the region where the cap stops.}
\label{f-6-5-15} 
\vspace{1.cm}
\centerline{\psfig{file=sti-6-5-16.ps,height=3.2truein}}
\caption[f-6-5-16]
{Stresses in the toroidal field coil for the fourth model with 450~MPa maximum
stress in the inboard side to the coil, top and bottom.}
\label{f-6-5-16} 
\end{figure}

\begin{figure}[p] 
\centerline{\psfig{file=sti-6-5-17.ps,height=3.1truein}}
\caption[f-6-5-17]
{Stresses in the external coil structure for the fourth model with 700~MPa
maximum stress in the region where the cap stops being toroidally continuous.}
\label{f-6-5-17} 
\vspace{1.cm}
\centerline{\psfig{file=sti-6-5-18.ps,height=3.1truein}}
\caption[f-6-5-18]
{Displacement in the external coil structure for the fourth model with 0.05-m
maximum displacement in the region where the cap stops being toroidally
continuous.}
\label{f-6-5-18} 
\end{figure}

\clearpage

\section{SUMMARY AND CONCLUSIONS}
\label{s-6-6}

Three major classes of engineering design options have been examined to
determine if they meet the Demo requirements and to characterize the remaining
technical issues.  The design options were categorized according to the primary
in-vessel structural material -- ferritic steel, vanadium alloy, or SiC
composite.

Silicon-carbide composite has a unique potential for safety and high
performance, but the database requires substantial improvement in order to
identify and develop a satisfactory material composition.  Significant
improvements are needed in the basic properties, and several key material
issues such as thermal conductivity improvement, joining and hermeticity must
be resolved.  It was judged premature to adopt such a material for Demo, which
was assumed to have a relatively near-term schedule given the small R\&D effort
on this material in U.S.

Ferritic steel has the largest database, and hence the smallest uncertainty in 
its performance.  However, as with all materials, behavior after long-term 
operation in a fusion environment is highly uncertain.  The features of 
ferritic steel which cause the greatest concern are its restricted temperature 
window, which limits the maximum achievable thermal conversion efficiency, and 
its loss of ductility.  It is not clear how one would operate a high-power 
device with potentially brittle materials.  Economic studies indicate that the 
lower cost of steel (as compared with vanadium alloy) does not completely 
offset the effects of lower power density and lower thermal conversion 
efficiency.

Vanadium offers significant advantages in its high temperature and high 
thermal performance capability, and its low activation.  The combination of V 
and Li is particularly unique in material chemistry, offering good 
compatibility up to very high temperatures.  The primary concerns with 
vanadium alloy, besides the obvious concerns over their behavior in a fusion 
neutron environment, are its high cost and liquid metal MHD effects (including 
the need for and feasibility of insulating coatings).  Minimization of 
vanadium in the shield and external systems, together with lifetime extension 
and/or recycling is an important cost-reduction strategy.  The absence of an 
established industrial base is also an important consideration, but not one of 
the top-level requirements as elaborated earlier.  Due to its greater ultimate 
potential for attractive commercial power plants and a development path which 
appears practical within a 25-year time-frame, the combination of Li 
breeder/coolant and vanadium-alloy structure was chosen as the first design 
concept to undergo full system design and analysis in Starlite.

Superconducting magnet design options also were assessed.  Several fault
situations were analyzed in order to assess the necessary modifications in the
design.  Maintenance and configurational issues place demanding constraints on
the position and shape of TF coils.  In one scheme, proposed for \AI, one TF
coil is replaced along with its fusion core module.  This step requires a
separate dewar/cryostat for each coil.  Two schemes for dewar/cryostat system
for the toroidal field system were investigated to assess the possibility of
removing a TF coil during the maintenance operation.  It was found that a
separate dewar/cryostat requires substantial additional insulation in the
inboard leg of the TF coils, making this option costly.  In another maintenance
scheme, proposed for \AIV and PUSLAR, an entire sector is remove radially in
between the TF coils (the outer leg of TF coils has to be extended outward).
Several structural designs were investigated to assess the feasibility of
having large access ports in the outboard side.  Preliminary analysis indicates
TF coil structure can be designed to allow for large access ports.

\clearpage
\begin{thebibliography}{99}

\bibitem{BCSS} \BCSS.

\bibitem{ARIES1} \ARIESI.

\bibitem{ARIES4} \ARIESIV.

\bibitem{Klu94} R.~L.~Klueh, {\it Proc. IEA Working Group Mtg.
Ferritic/Martensitic Steels,} Sun Valley, Idaho (June 24, 1994).

\bibitem{Dal95}
M.~Dalle Donne, L.~Anzidei, H.~Kwast, F.~Moons, and E.~Proust, 
``Status of EC Solid Breeder-blanket Designs and R\&D for DEMO 
Fusion Reactors,'' {\it Fusion Eng.\ Des.,} {\bf 27} (1995) 319-336.

\bibitem{Gia95}
L.~Giancarli, G.~Benamati, S.~Malang, A.~Perujo, E.~Proust, and J.~Reimann, 
``Overview of E.U.\ Activities on DEMO Liquid Metal Breeder Blankets,''
{\it Fusion Eng.\ Des.,} {\bf 27} (1995) 337-352.

\bibitem{Klu95} R.~L.~Klueh and D.~J.~Alexander, ``Embrittlement of Cr-Mo
Steels after Low Fluence Irradiation in HFIR,'' {\it J.\ Nucl.\ Mater.,} {\bf
218} (1995) 151-160.

\bibitem{Klu94a} R.~L.~Klueh, Ji-Jung Kai. and D.~J.~Alexander,
``Microstructure-Mechanical Properties Correlations of Irradiated Conventional
and Reduced-Activation Martensitic Steels,'' in {\it Fusion Materials
Semiannual Progress Report for the Period Ending March 31, 1994,} U.S.-D.O.E.\
report \hbox{DOE/ER-0313/16} (1994).

\bibitem{eluebaly}L. A. El-Guebaly, C. G. Bathke, and the ARIES Team,
``Cost-Effective Shield Designs for Environmentally Attractive Fusion Power
Plants,'' {\it Fusion Tech.} {\bf 30}, Dec. 1996, 1058.

\bibitem{STARFIRE} \STARFIRE.

\bibitem{ONEDANT} R.~O'Dell \ea, ``User's Manual for ONEDANT: A Code Package
for One-Dimensional, Diffusion-Accelerated, Neutral Particle Transport,'' Los
Alamos National Laboratory report, \hbox{LA-9184-M} (1982).

\bibitem{DKR-ICF}
D.~L.~Henderson and O.~Yasar, ``DKR-ICF:  A Radioactivity and Dose 
Rate Calculation Code Package,'' University of Wisconsin Fusion 
Technology Institute report \hbox{UWFDM-714} (1987).

\bibitem{REAC3}
F.~M.~Mann and D.~E.~Lessor, ``REAC*3 Nuclear Data Libraries,'' 
{\it Proc. Int. Conf. on Fusion Nuclear Data,} Forschungszentrum Julich, 
Germany  (May 13-17, 1991).

\bibitem{10CFR20}
``Standards for Protection Against Radiation,'' Nuclear Regulatory 
Commission, 10CFR part 20 (1975).

\bibitem{WDR}
H.~Y.~Khater, ``WDR:  A Code for Rad-waste Classification,'' unpublished.

\bibitem{10CFR61}
``Licensing Requirements for Land Disposal of Radioactive Waste,''
 Nuclear Regulatory Commission, 10CFR part 61, {\it Federal Register},
{\bf FR 47} (1982) 57446.

\bibitem{Fet90} S.~Fetter, E.~T.~Cheng and F.~M.~Mann, ``Long Term Radioactive
Waste from Fusion Reactors: Part II,'' {\it Fusion Eng.\ Des.,} {\bf 13} (1990)
239.

\bibitem{Lec93} T.~Lechtenberg, C.~Dahms, and H.~Attaya, ``An Assessment of
Magnetic Effects in Ferromagnetic Martensitic Steels for Use in Fusion
Machines,'' {\it Proc. Topical Conf. Ferritic Alloys for use in Nuclear Energy
Technologies,} Snowbird, Utah (June 19-23, 1983) 179.

\bibitem{TNS} ``GA TNS Project,'' General Atomics report GA-A15100 (1978).

\bibitem{Bec80} W.~R.~Becraft and P.~J.~Reardon, ``Scope and Status of the
U.S.A.\ Engineering Test Facility Including Relevant TFTR Research and
Development,'' \IAEA{8} 563.

\bibitem{UWMAK1} \UWMAKI.

\bibitem{UWMAK3} \UWMAKIII.

\bibitem{Ehs87} D.~Ehst \ea, ``Tokamak Power Systems Study-FY 1986: A Second
Stability Power Reactor,'' Argonne national Laboratory report
\hbox{ANL/FPP/86-1} (1987).

\bibitem{TITAN} \TITAN.

\bibitem{Won94} Wong, C.~P.~C. \ea, ``A Robust Helium-cooled Shield/Blanket
Design for ITER,'' \IEEE{15}.

\bibitem{Sal95}
E.~Salpietro, ``Reactor models for Safety and Environmental 
Assessment of Fusion Power (SEAFP),''  {\it Proc. U.S./Japan Workshop 
on Fusion Power Plants,} Kyoto University, Japan (Mar.\ 13-17, 1995)
Japan Atomic Energy Research Institute report 07-100 (1995).

\bibitem{Dal94} M.~Dalle Donne \ea, ``European DEMO BOT Solid Breeder
Blanket,'' Kernforschungszentrum, Karlsruhe report KfK5429 (1994).

\bibitem{Nat95} K.~Natesan, C.~B.~Reed, and R.~F.~Mattas, {\it Fusion Eng.\
Des.,} {\bf 27} (1995).

\bibitem{Tob86} A.~Tobin and G.~Busch, {\it J.\ Nucl.\ Mater.,} {\bf 141-143}
(1986) 604.

\bibitem{Cot72} C.~F.~A.~Cotton and G.~Wilkenson, {\it Advanced Inorganic
Chemistry,} 3rd Edition, Interscience Publishers (1972).

\bibitem{Sat94} M.~Satou, ``Status of V-Ti-Cr-Si (Al, Y) alloy development,''
{\it Proc.  IEA Workshop Vanadium Alloys for Fusion,} Salem, OR (June 1994).

\bibitem{Mar94} Marshall, T.~D.~and Cadwallader, L.C., ``In-Vessel ITER Tubing
Failure Rates for Selected Materials and Coolants,'' Idaho National Engineering
Laboratory report \hbox{EGG-FSP-10928} (1994).

\bibitem{Won93} C.~P.~C.~Wong, ``Analysis Supporting BOTOS Helium Related
Coolant Tasks, Revision 1, with minor changes in attachment 2.2,'' General
Atomic memo (Aug.\ 10, 1993) unpublished.

\bibitem{Gar68} R.~A.~Gardner, ``Laminar Pipe Flow In a Transverse Magnetic
Field with Heat Transfer,'' {\it Int.\ J.\ Heat Mass Transfer,} {\bf 11} (1968)
1081-1083.

\bibitem{Dal79} M.~Dalle Donne, ``Comparison of the Performance of 1000 MWe
GCFR's with Two and Three-dimensional Roughnesses on the Fuel Rods,'' {\it
Proc.  5th GCFR Specialist Mtg.,} Wurenlingen, Switzerland (May 14-16, 1979),
OECD-NEA Coordinating Group on Gas-Cooled Fast Reactor Development (1979).

\bibitem{Kak87} S.~Kakac \ea, {\it Handbook of Single-Phase Convective Heat
Transfer,} John Wiley \& Son (1987).

\bibitem{ITER-TAC4} ``Detail of the ITER Outline Design Report, The ITER
Machine,'' ITER TAC-4 Meeting (Jan.\ 10-12, 1994).

\bibitem{Bro94} J.~Brooks, Argonne national Laboratory, personal communication
(April 1994).

\bibitem{ITER-DIV} ``U.S.\ Design Support Study for the ITER Divertor,'' ITER
report \hbox{ITER/US/93/IV-PF-3} (Oct.\ 1993).

\bibitem{Wam95}
W.~R.~Wampler, ``Post Exposure RBS Analysis of DiMES sample 70,'' 
personal communication (Jan.\ 1995).

\bibitem{Grexx} D.A.~Greene, D.C.~Gaubatz and K.R.~McMurtrie, ``Leak Location
in a Double Wall Tube Steam Generator,'' {\it Proc.\ 2nd Int.\ Conf.\ Liquid
Metal Technology in Energy Production,} Richland, Washington.

\bibitem{Smi94}
D.~L.~Smith \ea, ``U.S.\ Contribution, 1994 Summary Report, 
Task T12:  Compatibility and Irradiation Testing of Vanadium Alloys,''  
ITER report ITER/US/95/IV MAT 10 (1994).

\bibitem{Tav95} A.~A.~Tavassoli, ``Assessment of Austenitic Stainless Steels,''
{\it Fusion Eng.\ Des.,} {\bf 29} (1995) 371-390.

\bibitem{War95} J.~Wareing and A.~A.~Tavassoli, ``Assessment of Martensitic
Steels for Advanced Fusion Reactors,'' {\it Proc. SMiRT-13,} Porto Alegre,
Brazil (Aug.~13-18, 1995).

\bibitem{Gar91} F.~A.~Garner and R.~J.~Puigh, ``Irradiation Creep and Swelling
of the Fusion Heats of PCA, \HT and 9Cr-1Mo Irradiated to High Neutron
Fluences,'' {\it J.~Nucl.\ Mater.,} {\bf 179-181} (1991) 577-580.

\bibitem{Smi94a} D.~L.~Smith, R.~F.~Mattas, and M.~C.~Billone, ``Fusion Reactor
Materials,'' in B.~R.~T.~Frost (Ed.), {\it Nuclear Materials,} Vol. 10B,
Chap. 10 (1994) pp 243-340.

\bibitem{Col95}
A.~P.~Colleraine, General Atomics, personal communication (July 1995).

\bibitem{But94} G.~J.~Butterworth, K.~A.~McCarthy, G.~R.~Smolik and
C.B.A.~Forty, ``Safety and Environmental Aspects of Vanadium Alloys,'' {\it J.\
Nuc.\ Mater.,} {\bf 212-215} (1994) 667-672.

\bibitem{Mal95} S.~Malang, H.~U.~Borgstedt, E.~H.~Farnum, K.~Natesan, and
I.~V.~Vitkovski, ``Development of insulating coatings for liquid metal
blankets,'' {\it Fusion Eng. Des.,} {\bf 27} (1995) 570-586.

\bibitem{mattas}
J. H. Park and T. F. Kassner, ``CaO Insulator Coating and Self-Healing of
Defects on V-Cr-Ti Alloys in Liquid Lithium System,''
{\it J. Nucl. Mater} {\bf 233-237}, 476 (1996).

\bibitem{Mor95} H.~Moriyama, S.~Tanaka, D.K.Sze,J.Riemann, and A Terlain,
``Tritium Recovery from Liquid Metals,`` {\it Fusion Eng. Des.,} {\bf 28}
(1995) 226-239.

\bibitem{Mar75} V.~A.~Maroni, R.~D.~Wolson, and G.~G.~Staahl, ``Some
Preliminary Considerations of a Molten Salt Extraction Process to Remove
Tritium from Liquid Lithium Fusion Reactor Blankets,'' {\it Nucl. Technol.,}
{\bf 25} (1975) 83.

\bibitem{Rad95} A.~Radovinsky, Massachusetts Institute of Technology, Plasma
Fusion Center, private communication (1995).

\bibitem{SOLDESIGN} R.~D.~Pillsbury, ``SOLDESIGN User's Manual,''
Massachusetts Institute of Technology report \hbox{PFC-RR-65} (1991).

\bibitem{Fen93} J.~Feng, Massachusetts Institute of Technology, Plasma Fusion
Center, private communication (1993).

\bibitem{Sla80} G.~A.~Slack, ``Thermal Conductivity of Ice,'' {\it Phys.
Rev.\ B,} {\bf 22} (1980) 3065.

\end{thebibliography}

\end{document}

\bibitem{r55}
T.~J.~Dolan and G.~J.~Butterworth, ``Vanadium Recycling for Fusion Reactors,'' 
EGG-FSP-10378 (1994); and T.~J.~Dolan and G.~J.~Butterworth, 
``Vanadium Recycling,'' Fusion Technology (1994).

\bibitem{r56}
G.~J.~Butterworth, ``Recycling of vanadium-based alloys,''  
Proceedings of the Workshop on Low Activation Materials, Culham 
Laboratory, UK, 8-12 April 1991, Commission of the European 
Communities, EUR FU BRU/XII-6/91-MATIA 16, Volume II.