Professional Documents
Culture Documents
A Review of Lifetime Analyses For Tokamaks
A Review of Lifetime Analyses For Tokamaks
llfbtlme offirstwalls
analyses
JournalofNuclear
Materials
85 & 86 (1979)135-145
0 North-Holland
Publishing
Company
S. D. HARKNESS(1)and B. CRAMER(')
System studies have shown that economic fusion power can best be achieved from the use of long
lived components. The stresses generated in a first wall module are a complex function of its
geometry, the chosen structuralmaterial and the tokamak burn cycle characteristics. A means of
applying ASME Code Case 1592 to preliminarydesign has been established. Methods of incorporating
some of the material property changes expected from irradiationare discussed. Cyclic stresses
imposed by tokamak operationare expected to cause fatigue related propertiesto govern the life
of the structure. Stress assistedbubble growth is also discussed. This may be the critical
mechanism in establishingthe creep rupture life of a fusion first wall component.
1. INTRODUCTION
135
136 S.D.Harkness and 3. Cramer i Review of lifetime analyses for tokamaks
a certain point in time. The number of failures conductivityof the wall. The actual wall
to be expected in the system will be the product temperaturewill, of course, be higher since
of the number of componentstimes the proba- the convectiveheat transferbetween the wall
bility of failure of one componentand may have and the coolant stream will never be perfect.
to include an environmentalweighting factor. If one assumes as a basis of comparisona
Most fusion reactor system designs involve coolant stream temperatureof 225*C, the re-
hundreds or even thousandsof first wall modules. lationshipbetween temperatureand wall thick-
For a short lived design the number of projected ness for three candidatematerials are as pre-
component failureswould be much higher than sented in Table 1. While each of these three
for a long lived componentover the lifetime candidates,Type 316 stainlesssteel, a stabi-
of the reactor plant. lized 9 Cr ferritic stainless,and a titanium
allay (Ti-6Al-4V)have an upper temperature
Thus considerationsof the economicsof fusion limit of about 5OO*C in a fusion reactor environ-
power indicate the incentiveto achieve an ex- ment, It can be seen that widely differentwall
pected componentlifetime equal to that of the loading limits result strictly from the radial
reactor plant or ti30 years, To be useful, this temperaturegradient. Based on this alone,
life must be achieved for wall loadings of from titanium alloys are restrictedto service at
2-3 MW/m2, thereby creatingan overall goal of wall loadingsbelow % 4 MW/m2 if a wall 3 mm or
as high as 90 MW-yrfm2. greater in thicknessmust be used.
2. LIFETIME ESTIMATES
Table 1. Relative TemperatureLimitationsof
Lifetime estimatesare dependenton criteria Different StructuralAlloys.
developed in three categories. The first of
these includes those processes that actually
lead to failure of the component,either
through leakage of the coolant or by a com- AT Across
plete failure of an element. Creep-fatigue 1 cm wall,
and flaw propagationwould fall into this First Wall
category. Alloy I l. Mw$i?, OC
The second category involves limits on di-
mensional change that would cause some planned
operation to become impossible. Swelling and
creep are the main considerationsin this cate- $16 108
gory. Setting limits for allowabledimensional 192
change is design dependentand can only be done ri-6Al-4V
after all planned maintenanceoperationsare II?erritic,9-Cr 86
well established. A more stringentrestriction L
1 -200
0 10 20 30 40
Yield
Thermal Stress at
Strength
Structural at 500°C Wall Loading MWW/m2
Material MPa (psi) 1 3 1 5
Fig. 2. Longitudinalcompressionstress at
outer wall surface during burn period. t
I I I I I I I I I
*
2.2
MY/m.
TM,=503
'C,316
STAINLESS STEEL
- 30
also result in significantstresses. Recent 200 -
--- THERMOELASTIC YOOEL
calculations[7] for a 9-Cr stabilized ferritic L
s -
_--_ THEAYOELASTIC YODEL INCLUDINB ANELASTIC 2
stainless steel structure indicate pressuresof EFFECT AN0 STRESS RELAXATION
2,1 atm will be imposed if this material is i d
used.
A possible source of cyclic primary stresses
are pressure fluctuationsdue to turbulencein
the coolant. These could contributeto fatigue
failure in first wall structures. In addition,
plasma disruptionsmay cause electricalcurrents
to flow in first wall structuresresulting in
internal forces. These must be combinedwith
other primary loads in determiningstructural
thicknesses. These dynamic forces and their
resultingeffects on stress have been shown to I I I I I I I I I
have a significanteffect on structuraldesign 0 IO 20 30 40 SO 60 70 60 SO 100
and on resulting component life [8]. Because TIME. I
these disruptionsmay occur frequently (assumed
once each 10 cycles in Ref. [El), they may have
to be considereda normal loading condition.
Fig. 3. A calculationof the thermal stresses
In addition to these primary stresses, at the at the plasma side of a 3 aaathick, hemis-
onset of operation the incident electromagnetic phericallycapped first wall module during
and particle fluxes that comprise Q 25% of the a 60 second burn cycle with a 15 second dwell
total fusion energy cause large compressive period.
thermal stresses to be generated in the first
138 S.D. Harkness and B. Cramer /Review of lifetime analyses for tokamaks
0 IO 20 30 40 50 60 70 80
CYCLE TIME, s
2.0
t- \ / -+.o
Fig. 8. *The first wall surface temperatureas
a functionof time for a 60 second burn pulse
and a 15 second dwell period,
in the above relationship. This value of wall in a wall of 0.25 cm thicknessresulted in the
thicknesswas then used in a fatigue based informationpresentedas Fig. 11 [19]. For
equation these conditions,a significantimprovementin
u&W
the crack propagationpropertiesare necessary
Et (0 =- 2k(l-w) just to achieve a 10 MW-yr/m2 life at a wall
loading of 2 HW/m2. Based on this calculation,
only low stress, low temperaturecombinations
where Q(L) is the code given allowable strain
would allow the achievementof the minimal 10
to achieve the desired lifetime,W is the ther-
MW-yr/m2.
mal wall loading (equals* 25% of the neutron
wall loading),a is the thermal expansionco- In a study of an Inconel 718 first wall
efficient,k is the thermal conductivity,v is module, Kearney, et al. [5] found that wall
Poisson's ratio, and 6 is the wall thickness. loadingswere limited to less than Q 2.2
MW/m2 from their impositionof a 6OO'C maxi-
In the second step of the analysis, deform&ion mum temperature. For a neutron wall loading
limits on thermal ratchetingand thermal creep of 2 HWfm2, it was found essential to detect
were examined based on sn approach for analyzing all flaws larger than 0.75 mm in depth in
the effects of ratchetingin a thermallycycled order to achieve acceptablelifetimes (> 10
system developedby Bree [Ml. When Youngdahl MW-yr/m2) for their 5 rm~thick structure.
and Smith extend their analysis to cover
ratcheting,the allowable design point drops u
from 1.3 to 0.9 HWfm2 (i.e. from point A to
point B in Fig. 10) where the fatigue curve
plotted in Fig. 10 is taken from Fig. 9. Again
the results of includinga temperaturelimi-
7
tation reduces the attractivenessof using a
low structuraltemperaturedesign. @
I.1
-#
lo- .s -
w- .4_
-5 -
-6-
A&
rtr
AK--psIJIR
Wall Thickness, 6, mm
testing systems are shown in Fig. 12 for a wall Code Case requirements were addressed from the
thickness of 0.25 cm. If a scratch depth of viewpoint of what additional conservatisms
only 0.025 cm (10% of the wall thickness) is should be included when designing with a low
assumed, a substantial improvement is still re- (where the definition of low changes with the
quired in crack propagation properties if eco- requirement) ductility material. A major
nomically interesting wall lifetimes are to be assumption made in this work is that swelling
achieved. and irradiation creep do not inherently damage
the material.
a
&Gq t
For example, in setting the criterion for
the primary membrane stress intensity for un-
irradiated materials, the level must be kept
below the yield stress, S , while for irradiated
u material whose ductility {as dropped below 5%
(or as can be shown to be equivalent, whose
yield strength to ultimate strength ratio has
exceeded 0.6), the allowable level becomes 0.6
of the ultimate strength.
When including the effect of ductility loss on
the use of the Bree diagram to protect against
ratcheting, the authors conclude that it must
be amended when the uniform elongation is re-
duced to approximately 1% for an austenitic
stainless steel. At this point the work of
Leckie and Ponter [23] is used to modify the
boundaries in the Bree diagram. These authors
have shown that ratcheting strains will be
small when creep is significant if loads are
kept less than n/n+1 of the elastic-plastic
shakedown loads where n is the exponent in
the creep rate relation E, = AU". If the thermal
creep expression is used, then n's of 4-6 are
appropriate and relatively small movements of
the Bree diagram boundaries result. If the ex-
AK - psi'_Ji;; pression for radiation enhanced creep is used,
then an n value near 1 should be used and much
larger changes in the Bree boundaries are re-
quired. The use of any thermal creep related
deformation laws for the prediction of per-
Fig. 12. Effect of initial flaw size on re- formance of a component during neutron radiation
quired flaw rowth rate properties to achieve may at first seem questionable since, in
a 10 MW-yr/m !! life. the temperature range of interest, the time de-
pendent deformation mechanisms for the matrix
From this discussion, it is evident that of the grains of the structural alloy are not
flaw propagation must be included in any life- those that operate during thermal creep. In
time analysis of a fusion device. This should other words, cFeep during irradiation is not a
be done after the structure has been found superposition of a radiation enhanced process
acceptable in an analysis such as outlined by and a thermal one. However, the importance of
Youngdahl and Smith. including a limit on rhtcheting has more to do
with avoiding conditions where cracks are ini-
With this analysis framework in mind, the next
tiated within the material than on the absolute
step is to extrapolate the available data to amount of deformation that occurs. Therefore,
conditions presented by the fusion environment. what one would really like to know is the amount
of grain boundary sliding that occurs for some
7. EXTRAPOLATION OF DATA given amount of total strain. If one assumes
that the grain boundary sliding process is
The high energy neutron bombardment character- little affected by radiation, then the calcu-
istic of a fusion device will lead to major lation of a nonexistent amount of thermal creep
microstructural changes [20,21]. These effects strain may serve as an adequate estimate of the
are best documented for the austenitic stainless level of the grain boundary sliding (some small
steels where swelling and ductility loss are fraction of the actual calculated creep strain)
perhaps the most important property changes that can be incurred without crack initiation.
with fluence. An important paper 1221 has
recently been presented that deals with the For this reason, the suggestion by Nelson,
difficult problem of how to incorporate the et al., to use a total strain limit that is
based on a summation of the ratio of the plastic
ductility loss into a design approach. In
strain, 6~p, to the plastic strain limit, E:,
the referenced approach, each of the ASME 1592
S.D. Harkness and B. Cramer i Review of lifetime analyses for Tokamaks 143
AISI 300 series stainless steels welds. They ob- on the ASME Code Case 1592 due primarily
served large decreases in postirradiation frac- to the work of Youngdahl and Smith [17]
ture toughness in 308 stainless steel welds, and to Nelson and others [22].
based on fatigue precracked Charpy-V specimens
5. Fatigue and stress-to-rupture properties
and J-Integral assessment procedures. For may be critical to the achievement of long
EBR-II radiation exposures at temperatures lives by tokamak first wall modules.
above 370°C, fracture toughness values were
reduced to as little as 20% of unirradiated 6. When the propagation of undetected flaws
values. Reductions of this magnitude will have is included, much shorter lives result
an extremely detrimental effect on fusion reac- than predicted by the ASME based
tor first wall design and life. approach.