Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

MI Cable Conductor Studies

Download as pdf or txt
Download as pdf or txt
You are on page 1of 36

/ By acceptance of this article, the publisher o<

D,SCLA,MER - -, recipient acknowledges the U.S. Govo-nmc-


right to retain a non - exclusive, royni'-, •
license in and to any

Lifetime improvement of sheathed thermocouples for use in


a
high-temperature and thermal transient operations

R. W. McCulloch*
Oak Ridge National Laboratory, Oak Ridge, Tennessee 37830

J. E. Clift C

Y-12 Development Division, Oak Ridge, Tennessee 37830


DEf:2 00 8844
ABSTRACT

Premature failure of small-diameter, magnesium-oxide-insulated


sheathed thermocouples occurred when they were placed within nuclear
fuel tcl simulators (FRSs) to measure high temperatures and to follow
severe thermal transients encountered during simulation of nuclear re-
actor accidents in Oak Ridge National Laboratory (ORNL) thermal-hy-
draulic test facilities. Investigation of thermally cycled thermocou-
ples yielded three criteria for improvement of thermocouple lifetime:
(1) reduction of oxygen impurities prior to and during their fabrica-
tion, (2) refinement of thermoelement grain size during their fabrica-
tion, and (3) elimination of prestrain prior to use above their recrys-
tallization temperature. The first and third criteria were satisfied
by improved techniques of thermocouple assembly and by a recovery anneal
prior to thermocouple use.
Grain-size refinement to ASTM mirroETflin size 9 (3970 grains/mm2)
•as accomplished by adopting an appropriate annealing schedule and opti-
mizing the- drawing process. The general drawing equation was extended
to describe thermocouple internal stresses, process parameters were
evaluated and optimized, and a multiple-pass draw schedule was developed
that increased cold work (true strain) between anneals from 25 to 55To.
Maximum first-drrw cold work was limited to ~42% by frictional work that
was found to be 25% higher for thermocouple drawing than for equivalent
wire drawing. Second-draw cold work was limited to 10 to 20% by fric-
tion and high yield stress. More than 3500 ra of refined grain material
was produced; longest average lengths of -100 m were obtained with a
single-draw process of 40% cold work between anneals.

SUBJECT INDEX

1. Type K thermocouples - sheathed.


2. Type K thermocouples - fabrication.
3. Transient behavior of thermocouples.
a
Research sponsored by the Office of Research and Technology, U.S.
Department of Energy under contract ff-7405-eng-26 with the Union Car-
bide Corporation—Nuclear Division.

Of THIS WCWHI K
INTRODUCTION

Small-diameter sheathed thermocouples are the prin-


cipal means of temperature measurement in large out-of-
reactor thermal-hydraulic test facilities at ORNL and
other national laboratories, as well as at several pri-
vate companies conducting reactor safety experiments.
High temperatures (up to 1200°C) as well as temperature
transients (up to 50°C/s) are often required. The re-
liability of 0.51-mm, sheathed, insulated-junction type
K thermocouples used in these experiments has improved
significantly in the last few years and, in general,
has met program needs. However, requirements for small-
diameter thermocouples for use in the developing Gas-
Cooled Fast Breeder Reactor-Core Flow Test Loop (GCFR-
CFTL) program at ORNL necessitated a marked increase in
reliability for the CFTL to meet successfully its ob-
jectives.

CFTL FRS DESCRIPTION

The CFTL (Fig. 1) is a high-temperature, high-


pressure, fast-transient, out-of-reactor gas loop that
was designed to supply helium to a test bundle at ap-
propriate temperature and pressure conditions. The
test bundle consists of electrically heated FRSs that
are arranged to represent a segment of the GCFR core.
The CFTL is designed to accommodate rod bundles of 37
to 91 FRSs and to operate from ambient pressure to 10.3
MPa. from 300 to 137O°C, and at FRS power levels from
0 to 43 kW/rod (0 to 4 MW total power). The principal
objectives of the test program were the acquisition and
analysis of test data on (1) thermal and pressure-drop
characteristics, (2) design and safety margins, and
(3) friction and thermal mechanical behavior of test
bundles that simulate portions of GCFR fuel and control
assemblies during steady-state or transient operation.
Transient operation included prescribed variations
(with time) of test bundle power, helium pressure, and
coolant flow rate to simulate normal, upset, emergency,
and faulted accident transients. Data to be obtained
were primarily from the subject thermocouples, most of
which are integral components of the FRSs.
The design of the CFTL FRS is shown in Fig. 2.
The FRS cladding is prototypic of GCFR 316 stainless-
stecl-alloy fuel rod cladding and thus has the same
dimensions, configuration, and thermal and mechanical
characteristics. The length of the FRS is 2860 ran,
1200 mm of which is heated; the outside diameter is 8.0
mm. A length of 1220 mm of the outer surface, slightly
overlapping the heated length, is roughened to enhance
heat transfer to the helium. The FRS is reentrant :in
design; that is, both power terminals enter at a single
end. The outer terminal, a copper tube, is joined to a
tubular nickel conductor that, in turn, is joined to
one end of the variable-width heating element designed
to provide a cosine heat-flux distribution. The center
terminal is made of a solid copper rod that is joined
to a nickel conductor. This conductor, in turn, is
joined through an adapter to the other end of the heat-
ing element to complete the electric-current-carrying
circuit in the FRS.
Six type K thermocouples (four in earlier FRSs)
with insulated measuring (hot) junctions are installed
inside the FRS cladding. The sheaths of the thermocou-
ples run longitudinally along the inside surface of the
clad at 60° angular intervals. Each of the six thermo-
couple junctions is located axialiy with a tolerance of
+1 mm and is required to provide temperature informa-
tion over an extremely wide range. Table I summarizes
these requirements. The combinations of violent ther-
mal cycling, high temperatures, and long lifetime are
severe requirements for small-diameter type K thermo-
couples.

INITIAL FRS QUALIFICATION TESTS

To qualify the CFTL FRS and its thermocouples,


operation is at various steady-state conditions from
300 to 1100°C for 100 to 200 h, including 40 h at
1100°C. The CFTL FRS is then subjected to more than
2000 thermal cycles for a total operation time of about
1000 h. Thermal transients are normally from 1 to
S0°C/s, and the temperature changes during cycling are
from 150 to 600°C.
Initially, four FRSs were tested, into which a
total of 13 small-diameter insulated-junction thermo-
couples had been installed. About 1200 h of steady-
state and transient-cycling test? were conducted on the
FRSs, and over 1000 thermal transients were conducted
on one of them having 4 thermocouples. Most thermo-
couples failed in the initial steady-state portion of
the tests. Those thermocouples that did survive the
steady-state tests failed within a few hundred trans-
ient cycles. Failure in all cases was by development
of an open circuit in one or both of the thermoelements
at elevated temperatures. When the FRS temperature was
loitered, the thermocouples would again become opera-
tive. Photomicrographs of failed thermocouples showed
voids at the grain boundaries of both thermoelements,
and bcth thermocouples were extremely brittle.
These negative results prompted investigation into
the causes and mechanisms for failure. Investigations
were primarily directed toward lifetime improvement.
Previous work by Potts and McElroy1 and by Anderson and
Kollie2 showed that small-diameter thermocouples are
susceptible to metallurgical, chemical, and mechanical
changes and that thcie changes affect both lifetime and
accuracy.

THERMOCOUPLE INVESTIGATIONS
To understand the factors that contribute to the
premature failure of small-diameter thermocouples, both
their physical structure and metallurgical state must
be considered.
A thermocouple consists of two thermoelements, an
insulating material, and a metal sheath (Fig. 2 ) . One
thermoelement is Chromcl-P* (the K+ alloy), a nickel-
based alloy having chromium and silicon as the major
alloying elements. The other thermoelement is Alnmel*
(the K- alloy), a nickel-based alloy containing alumi-
num, manganese, silicon, and cobalt as major alloying
elements. The final average prain size of both thermo-
element.-; was specified for CFTL use to be ASTM micro-
grain size 7 (9"2 graJ«>s/ m m 2 ) or smaller. Thermoele-
ment wires are contained in a magnesium oxide insula-
tion also having a particle size of APTM micrograin
size 7 or smaller. The sheath is Irconel 600,'' a
nickel-based alloy. The thermoelenient wire diameter
and the sheath thickness are O.O7o to 0.089 mm. Thus,
the thermoelement grain size ar.d wire diameter are such
that as few as four to six grains traverse the entire
thermoelement cross section. Groin boundaries repre-
sent material discontinuities and, at high tempera-
tures, arc relatively weak compared with the rest of
the material; 3 therefore, regions along the thermoele-
ment with large grains are weak points within the
structure.
Another closely related factor influencing life-
time is thermal stress arising from differential ther-
mal expansion between the thermocouple components and
other FRS components. Figure 3 (Ref. 4) gives a com-
parison of the thermal expansion of the 316 stainless
steel used as the FRS cladding, the Inconel 600 thermo-
couple sheath material, and the Chromel and Alumel
thermoelement materials. At 1000°C, the 316 stainless
steel clad expanded 2OTo more than the Chromel ther-
moelement and 3OTo more than the Alumel thermoelement.
The thermocouple is intimately associated with the much
more massive FRS clad, and both thermoelements are put

•Trade narc of the lloskins Manufacturing Company.


^Trade name of the International Nickel Company.
into tension as temperatures are increased. These ten-
sile forces, imposed on grains only slightly smaller
than the thermoelement lead wires, are major contribu-
tors to premature thermocouple failure.
A third contributing factor is that of oxidation.
Spooner and Thomas, 5 operating bare Chromel and Alnmel
wires in various environments, found that oxidation of
both Chromel and Alumel thernoelements resulted in em-
brittlement and early failure, especially when the
thermocouples were operated in the temperature range
of 81" to iO38°C.
A final factor that has been found to make thermo-
couples susceptible to both early failure and tempera-
tuie measurement errors is residual cold work, or pre—
strain, in the thermoelements prior to their use at
temperatures above their recrystallization tempera-
ture.
Potts and McElroy 1 have shown the importance of a
fully recxystallized microstructure, free of material
cold work, as the final fabrication stop for type K
thermoelements. Cold working changes the Seebeck coef-
ficient of the material, and, during use in an VMS, the
cold worked material will be annealed, again changing
the Seebeck coefficient. Changes in the Seebeck coef-
ficient of a thermocouple cause temperature measurement
errors.
Jn addition to causing errors, small amounts of
cold Mfu/ti. have been shown to cause relatively large in-
creases in grain size, possibly by a factor of 2 or 3
(Ref. 6 ) , when a material recrystallizes. Figure 4
illustrates the effect of cold work on recrysta11ized
grain size. With thermoelement grain diameter already
not much smaller than the wire diameter, low levels of
prestrain clearly should be avoided, if only to pre-
serve a small grain size following recrystallization.
Anderson' conducted tests to determine the com-
bined effects of grain size, prestrain, and differen-
tial thermal expansion on thermocouple lifetime. Ther-
mocouple specimens from two manufacturers were used.
The first specimen had a grain size of ASTM 7 to 8, and
the second had a grain size of ASTM 6 to 7. Test sam-
ples were prepared by collapsing 1.5-mm-diam stainless
steel tubing over a length of 0.5-mm-diam thermocouple
material. The clamping of the collapsed tubing to the
outside diameter of the thermocouple sample was in-
tended to simulate the strong mechanical connection
that exists between the FRS cladding and the internal
thermocouples. The collapsed tubing was electrically
heated through various thermal cycles, and the loop
resistance of the thermocouples was monitored.
Those specimens with the smaller grain size (ASTM
7 to 8) were found to endure about 5000 cycles before
failure (open circuit and loss of electromotive force),
or 2 to 3 times the number imposed on those with the
larger grain size before failure. Also, prestraining
the samples 5% prior to thermal cycling reduced their
lifetime by a factor of 2.
Microstructures of most of the test samples showed
void formation at the grain boundaries. Anderson at-
tributed these voids to aluminum and chromium oxides
that apparently formed at the respective grain bound-
aries and were removed by metallograpitic sample prepa-
ration. In both Anderson's investigations and investi-
gations reported here, samples with voids were ex-
tremely brittle. The brittleness of Anderson's samples
was attributed by Ludwig 1 to the brittle oxides that
formed at the grain boundaries.
A quite different explanation of grain boundary
voids has been offered by Dyson and Henn' and by Dyson
and Rodgers. 1 0 In a study of a nickel-based alloy,
they found that prcstrain introduces microcracks that
nucleate to form grain boundary voids when the material
is subsequently heated above its rccrystallization tem-
perature. Further, the effect of stress is to enhance
the formation of voids. i'restrain values of about 2%
reduced the elon b ation—to—fracture measurements on sam-
ples by a factor of 2, indicating that even small val-
ues of prestrain result in grain boundary voids.
Although not specified, most thermocouples are
received with a residual cold work of 1 to 3% (Ref. 1 ) ,
presumably to adjust the Seebcck coefficient to meet
the calibration curve. Further, as part of their »s-
sembly into an FRS, the thermocouples are strained an
additional 0.5%. Thermocouples assembled into earlier,
swaged FRSs were prestraincd an additional 4 to 5% by
the swaging process. Thus, the thermocouples in a
swaged FRS (like those initially tested) have a prc-
strain of 5.5 to 8.5%, and those in a nonswaged FRS
have a prestrain of 1.5 to 3.59i>.
As the effects of prcstrain were surfacing, the
design of the CFTL FRS was csolving toward use of the
nonswaged fabrication technique, which, by its nature,
reduces thermocouple prestrain to a level below that at
which large grain growth occurs (Fig. 4 ) . A period of
low-temperature operation at ~35O°C for 1 to 2 d had
also been incorporated as part of the operational se-
quence. After t'icse changes were adopted, lifetime
tests of thermocouples in nonswaged FRSs were extremely
successful, although there was evidence of decalibra-
tion. Eight nonswaged FRSs with 27 type K thermocou-
ples endured over 5,000 h of steady-state and tran-
sient testing, including over 36,000 thermal transient
cycles. No thermocouples failed prior to FRS failure.
even though 4 rods endured 4,000 to 10,000 transient
cycles, many as severe as 50°C/s.
Figure 5(a and b) compares photomicrographs of the
Alum el thermoelement microstructure from a r-wnged FRS
prototype with a nonswageu FRS prototype. Both FRSs
endured about 1000 h of stsady-state and transient op-
eration and over 2000 thermal cycles. Sample 1 (Fig.
5a) failed after ~200 h of steady-state operation and
400 thermal cycles, while sample 2 (Fig. 5b) dit not
fail. Sample 1 was extremely brittle, but sample 2 was
ductile even though the grains were relatively large.
Thus, while oxidation contributes to failure and
should be minimized, the major cause of premature fail-
ure of small-diaaieter thermocouples appears to be pre-
strain-induced grain growth and grain boundary void
growth or oxidation, causing reduced thermoelement
cross-section and grain boundary embrittlement.
From the preceding discussion, the thermocouples
apparently must have three characteristics if they are
to endure high temperature and transient operation:

1. The thermocouples must be fully annealed when re-


ceived. Any ensuj'.g cold work occuring during
th-iir assembly should not exceed 15> and, for opti-
mum results, should be removed prior to operating
the thermocouples at or above their rccrystalliza-
tion temperature.
2. There loust be a minimum oi oxygen and moisture as
well as hydroxides [sucn as M R ( 0 H ) 2 J or oxides
(such as iron oxide) in the thermocouple insula-
tion. This requirement will help to reduce the
potential Hoc oxidation of the chromium or alumi-
num in the thermoelements.
3. The final grain size of both thermoelement wires
must be very small, with a recommended ASTM micro-
grain size 9 (~397O grains/nun2 or 0.016-mm average
diametsr).

Figure 4 shows that either very little or rela-


tively large amounts of cold work are not detrimental
to thermocouple grain size following recrystallization,
while intermediate amounts can result in undesirable
grain growth. A structure free of cold work (to pre-
vent recrystallization during use) can be obtained by a
recovery anneal below the material recrystallization
temperature.* Therefore, the first characteristic men-
tioned previously can be obtained by specifying a re-
covery anneal as the last operation prior to thermocou-
ple use.

The second characteristic can be obtained by speci-


fying a 1000°C, <10-u vacuum bakeout of the insulators
prior to their assembly into the thermocouple sheath,
followed by assembly of the thermocouple bulk material
in an inert atmosphere. The final characteristic re-
sults from optimization of the thermocouple fabrication
process by careful annealing and by using large amounts
of cold work to refine final grain size. The next sec-
tions discuss the background for and the methods used
to accomplish this.

RECRYSTALL1ZATION AND DRAWING

Recrystallization

Material recrystallization is a nuclcation and


growth phenomenon, the completion of which is dependent
on annealing temperature, time, and cold work performed
before annealing, as well as on composition variations.
A schematic representation of the effect of prior cold
work and initial grain size before cold working on the
recrystallized grain size is shown in Fig. 4. TJ.e
schematic suggests that a large amount of cold work is
desired to refine grains by subsequent recrystalliza-
tion; sraall amounts of cold work must be avoided unless
they arc too small to initiate recrystallization.
Because it is a nucleation and growth process, the
amount of material transformed during recrystallization
follows a sigmoidal distribution with time.* The time
at temperature generally follows an Arhennius-type re-
lationship with the material activation energy as the
exponential variable.
If the grain size of two materials is to be mini-
mized and if they must be annealed at the same tempera-
ture (as is the case with the Chromel-Alumel system in
a thermocouple), then the amount of rccrystallization
of the material with the highest activation energy
(Chromel) determines the annealing time. Thus the ei-
tent of recrystallization necessary for subsequent cold
working of that material without fracture must be de-
termined. This amount of recrystallization will deter-
mine the annealing time for the system and, therefore,
the grain size of the material with the lowest activa-
tion energy (Alumel).

Drawing

Reference 11 provides an excellent review of draw-


ing theory, part of which is presented here. The gene-
ral equation for drawing is

o d = (1/n) / o d e , (1)
where

a, = draw stress,
d
a = material stress at e,
i\ = process efficiency,
e = true strain in drawing (cold work).

For materials that obey the strain-hardening power law,

o = Ke n , (2)

where K is a strength coefficient and n is the strain-


hardening exponent. In this case, Eq. (1), after inte-
gration, becomes

o = a e1/ri(n + 1) , (3)

and for a second-draw pass, becomes

c a
a. = 1/t, — _ (e2 - 6l) . (4)
61
2

where a is the flow stress of the material exiting the


die, and

e = ln(A 0 /A 1 ) , (5)

where Ao and At are the entering and exiting areas,


respectively.
Figure 6 shows typical material stress and strain
plots during steady-state drawing as described by Eq.
(3). The material flow stress a increases in accord-
ance with Eq. (2) and is a at the die exit. The mate-
rial is strained to e = In ho/Ax at the die exit, and
the draw stress a acts 011 the reduced, strain-hardened
material.
Figure 7 shows the drawing operation for a first
draw described by Eq. (3) for t) < 1 and i\ = 1 as well
as Eq. (2) for the material. The material is influ-
enced by the draw die as it passes through but is again
in pure tension, in a work-hardened state, as it exits.
It is normally work hardened enough so that it will
neck and break if it yields. To prevent this, a must
be less than o so that the drawing limit is reached
when a = a . This is shown in Fig. 7 as the intersec-
tion between the 0 and a curves. For this case Eq.
(3) becomes
[e ]. = n(n + 1) . (6)
max draw
For pulling in pure tension,

[e ] .. ~ n . (7)
max pull
Comparison of Eqs. (6) and (7) indicates that a maximum
additional 100% elongation is possible with drawing as
compared with pullJng in pure tension. The amount
achieved depends en n, the drawing efficiency. For a
second drawing pass, the material flow curve would
start at o , and, because the material is strain hard-
e
ened, the o, curve would rise more sharply so that
di
(e2 - e x ) < e,.
The drawing efficiency r\ has been shown to depend
only on the product of process frictional and redundant
work terms and is given a s 1 2

1/tl = f* , (8)

where the process frictional work term is

f = (1 + u cot a) (9)

and the redundant work term is


• = II + C(A - 1)] , (10)

where

C = a constant (0.12 for wire),


(i = coefficient of friction,
a = draw die scmiangle,
A = the deformation zone geometry factor - oVL
(Fig. 6).

From Eqs. (8) through (10) the process efficiency


depends only on A, u, and a. To maximize the total
cold work e between anneals, the product of the factors
containing a and A must be minimized. (The friction
coefficient u depends primarily on the lubricant. Many
factors go into its selection and, although same inves-
tigations were conducted, to further minimize n was
beyond the scope of this work.)
Process parameters a and A arc related by

n (1 + e" E/l )
a (11)
180 (1 - e~6)
The friction factor [Eq. (9)] asymptotically decreases
with increasing a whereas the redundant factor [Eq.
(10)] linearly increases with A so that their minima is
reached at some intermediate value of a and correspond-
ing A.

APPLICATION TO THERMOCOUPLE DRAWING - T1IE


INTERNAL STRESS FACTOR
The preceeding analysis is adequate to arrive at
optimum conditions for drawing of wire. For thermo-
couple drawing, however, drawability is normally lim-
ited by one of the internal thermoelements breaking
(open circuiting). The dialing process must, there-
fore, be optimized based on the cold work and stresses
occurring on these internal components. The friction
factor [Eq. (9)] results from die-to-thermocouple sur-
face forces; it should be independent of e and can only
influence the mean draw stress, not the radial stress
distribution across the thermocouple.
The redundant term, however, has as its basis the
occurrence of nonuniform die pressure and resultant in-
ternal stress and strain distributions as functions of
the deformation zone geometry factor A. Figure 8 sche-
matically represents stress and strain distributions as
they might appear across a thermocouple cross section
for A = 3. The strain plot was extrapolated from hard-
ness data 1 3 on lubricated wire for A = 6.9 and lubri-
cated strip for A = 4, both with a true strain or cold
work for e = 0.17. From the plot, cold work can be
seen to have a large variation across the cross section
and is at a minimum internally. The average strain 7
is increased above e for the nonuniform distribution by
the value of * given in Eq. (10). For A = 5, the aver-
age cold work on the thermoelements is only about 0.8 e
while the sheath incurs ~2 e. As A decreases towards
one, the strain distribution becomes uniform. The im-
plication is that low A drawing is more suitable to
thermocouples if internal cold work is to be maximized
as needed to refine thermoelement grain size.
The stress distribution of Fig. 8(c) further re-
inforces this notion. The distribution shown was de-
rived from Hill's11* slip-line fiejd analysis (for fric-
tionless plain-strain compression of a nonhardening
material) and from Backofen's 1 5 maximum stress distri-
bution analysis of Hill's data. Backofen showed that
the variation in the internal stress distribution cr
xx
from surface to center followed the same relationship
with A as did the ratio of die pressure to surface
yield stress (P/a ) reported by Hill. The maximum in
this internal stress variation is given by
a__ = 2 a X , (12)
xx
max y
where X is the deformation zone sMp-line apes asgle
and is 0 for A = 1 and 0.43 n for A = 8.75. By inter-
polation, the maximum centerline to surface stiess with
A is then

0_. = 0.35 a (A - 1) . (13)


" max y

The stress distribution shown in Fig. 8(b) for A = 3 is


constant from the surface inward for a distance L/2 =
Tj^/A (where L is as shown in Fig. 6 ) , then rise., ap-
proximately linearly to a maximum of ~0.7 a (a = 28 x
10 6 kg/m 2 ) at the center. The draw stress o is ~0.1
o larger because oj this distribution, and the Inter-
nal stress o,_ is
dE

°y = °dr

where

2 / A - 1V
a = 0.35 a (A - 1) l -r (15)
y y 3 V A
and tie internal stress factor y is

a
y = 1 + — . (16)

Equation (3) for the first draw in terms of the inter-


nal stresses now becomes

dEl x.r,

If there is to be a second draw of the same material,


the stress distribution of the first draw must be taken
into account. This distribution o is additive to the
xx
internal stress developed in the second draw. However,
for A n on the first draw (which should normally be
the case), a << a so that a = a and the internal
xx
xx xx Y
stress distribution can be accurately reprei.ented by
Oy . A second distribution described by Eq. (15) is
developed for the second draw, which results in ay .
The internal stress after the second draw is the sum of
these two stresses. Equation (4) for the second draw,
in terms of the internal stress, becomes

where y2 is now

Equations (3) and (4) for the draw stress and Eqs.
(17) and (18) for the internal stress together enable
complete modeling of the thermocouple drawing process.
In their application to thermocouple drawing, the as-
sumptioAs of nonstrain hardening and homogeneity must
be recognized although the ability of the model to pre-
dict and improve the process will ultimately determine
its usefulness. The strain-hardening assumption, al-
though intended for nonstrain-hardeniiig material, was
shown by Backofen to still provide a reasonable upper-
bound estimate of redundant strain properties of strain-
hardening materials. 1 * The effect of nonhomogeneity
will be discussed in the next section.
To examine the internal stress as the limiting
parameter in thermocouple drawing, the friction, redun-
dant, and internal stress factors in Kqs. (17) and (18)
must be calculated for various a and A. Figure 9 plots
the Ofy product of Eq. (17) as a function of a for va-
rious amounts of cold work (e). The deformation zone
geometry factor A was determined for each a using Eq.
(11). The draw stress o was determined using Eq. (3),
and K and n in Eq. (2) were determined to be 137.0 x
10' kg/m 2 and 0.28, respectively, from tensile tests of
Inconel-sheathed, type K thermocouples.
The $fy product acts as a multiplier to the useful
work in drawing /ade, which results in the internal
stress a . As the useful work increases (with e) the
efficiency n = l/<J>fy improves. Figure 9 and Eq. (3)
indicate that a first draw of c = 0.44, with a between
10 and 12° and A between 1.6 and 1.9, is optimum and
will give a a -to-a ratio of -0.7, adequate to ac-
count for process and model uncertainties.
Additional cold work bsfore annealing can be ob-
tained with a second draw. However, as Fig. 10 shows,
the cold work is less and process parameters are much
more critical. The internal stress, resulting from
redundant strain and characterized by A, dominates the
process for a above 3 or 4 ° , and friction dominates for
a at less than 3 ° . The more pronounced minima ar.J
larger <J- fy product for the second draw results from the
greatly increased yield stress canicti by strain harden-
ing during the first draw. 'flic yield sti"ss, initially
35 x 10' kg/m 2 , increases by a factor of 3 to 4 to o
for tiie second draw. From Eq. (15), this proportion-
ally increases the internal stress distribution with A.
It is therefore extremely important to draw with A
close to 1 on the second draw.

THERMOELEMENT GRAIN-SIZE REFINEMENT

The development to produce the refined-grain ther-


moelements was performed by ORNL during fabrication of
a quantity of bulk thermocouple material for ORNL by
Groth-Mazur, Inc., AddIson, Illinois. Using the actual
production equipment avoided the effects on the final
results of small variations in specific parameters such
as deformation, furnace temperatures, and temperature
gradients.
The fabrication process consisted of the following
steps. The magnesium oxide insu'ators were subjected
to a 1000°C vacuum bakcout, and all materials were
thoroughly cleaned. The thermoelement wires and nag-
nesiuri oxide insulators were then assembled into an
Inconcl 600 sheath ~5 mm in diameter The diameter of
this assembly was reduced to 0.51 mm by a multiple-draw
anneal cycle.
The development effort optimized both the drcwing
and the annealing processes to achieve the finest grain
size possible, consistent with reasonable production
times and materiel yields.
The new drawing and annealing schedule to obtain
grain refinement was determined from (1) the preceding
recrystal1ization and drawing theory, (2! litcrclurc
information and physical property tests on the thermo-
couple components, and (3) industrial experience. The
development consisted of three general tasks:

1. adjustment of annealing conditions to minimize


grain growth of the Aluncl thermoelement while
still obtaining lull recrystal1ization of the
Inconcl sheath and Chrcnel thernoclenent,
2. definition of the existing drawing process in terns
of Eqs. (3) and (4) and calculation of the friction
factor f and coefficient of friction u, and
3. determination of a new process to maximize the re-
duction between anneals by optimizing process pa-
rameters a and A in terns of Eqs. (17) and (18) and
by comparison of draw and material stresses o and
a .
e
Annealing conditions were determined by drawing
annealed samples of Chrome! and Alumcl wire, and of
assembled thermocouples, to 20 to 5C& cold work, fol-
lowed by annealing at several time-temperature condi-
tions. ElongatAon-to-failure tests, as well as aetal-
lographic and microhardness examination of annealed
samples, were then used. Optimum annealing conditions
were found to be 1025°C for 60 s for a l-rm-diam ther-
mocouple. The annealing temperature was then held
constant and the time varied to account for thermocou-
ple mass-per-unit-length changes with diameter. An-
Aealf.ng times ranged from 120 s at about 3-inn diameter
tc 30 s at 0.5-nm diameter.
The applicability of drawing Eqs. (3) and (4) to a
nonhomogeneous thermocouple wzs investigated by (1) con-
ducting tensile tests on Iuconcl, Ch'oncl, and Alunel
wires and on type K, Inconcl-sheathed thermocouples and
(2) conducting draw tests on samples of thermocouples
and the Chromel thermoelement. Figure 13 summarizes
the tensile tests. Comparison of plots of measurements
on the individual thermocouple components with the ther-
mocouple measurements indicates that magnesium oxide
insulation substantially affects the tensile character-
istics of the thermocouple even though it cannot itself
sustain a tensile stress.
Strain-hardening constants for Eq. (2) were ob-
tained by curve fitting using the thermocouple and
Chromel stress-strain plot, and these values were used
in Eqs. (3) and ( 4 ) . Draw stresses a, and o, were
di d*
measured with a standard spring balance placed between
the draw bench-puller jaws end the thermocouple or
Chromel wire being drawn. The friction factor f was
then calculated for each of the samples. The results
showed f to be -25% higher for thermocouple drawing
than for wire drawing. This apparently reflects the
work associated with movement of the insulation as the
thermocouple is reduced as well as relative movement of
insulation material and thermocouple metallic compo-
nents as the insulation transmits the draw stress to
the thermoelement during thermocouple reduction. The
insulation thus contributes to the work of deformation
without supporting a balancing tensile stress.
The increased friction factor for thermocouple
drawing serves to increase the draw stress for <a given
reduction or, equivalcntly, to reduce the maximum
elor ,*tion possible for • given draw stress.
Calculation of the coefficient of friction u was
accomplished by drawing thermocouple samples with ex-
isting dies (with a = 6°) from e = 10% to e = 3O7J while
measuring the draw stresses o, or o , . Strain-harden-
di d ?.
ing constants from the thermocouple plot of Fig. 11
wore used in Eqs. (3) and (4), and Eqs. (9) and (10)
were used to calculate the frictional and redundant
work terms, respectively.
Friction coefficient calculations from these mea-
surements at first exhibited large variations. Draw
dies that exhibited a 1arger-than-anticipated draw
stress were replaced with new dies. In some cases, as
much as a 4 0 % decrease in draw stress was obtained for
a new die under the exact drawing conditions imposed
on the old one. This was attributed to die wear and
variations in die bearing length B (Fig 6 ) . After die
replacements, u was measured to be 0.19 + 1 0 % for 90%
of the dies tested although variations up to 2 5 * oc-
curred.
The existing process consisted of drawing to 20 to
25% cold work, in cue step between anneals using draw
dies with a = 6 ° . During development, thicc iiew pro-
cesses were obtained. In the first, the existing first-
draw cold work (or slightly more) was retained, and a
second draw of about 15% was added. Existing dies with
a = 6° «cre used, in most cases, for both draws. In
the second process, new dies with larger semiangles
were designed, and the l'irst-draw cold work was ex-
tended to about 4 2 % . The final process combined the
4 0 % maximum first-draw cold work with 15° second-draw
cold work to obtain 55% cold work between anneals.
Optimization of the drawing and annealing schedule
required continuity checks and elongation-to-failure
tests of postanncalcd material. Excess stresses often
did not immediately result in thermoelement open cir-
cuits. However, elongation-to-break tensile tests of
postannealled thermocouples were excellent indicators
of damaged or borderline material. Material correctly
drawn and annealed experienced a true tensile elonga-
tion of 20 to 25% for both thermoelements and sheath,
and the failure occurred initially in the sheath. Dam-
aged material was characterized by initial failure of
the Chronic 1 thermoelement at a true strain of only 10
to 15% and, occasionally, failure of the Alumcl thermo-
element prior to that of the sheath as well. Material,
once damaged, performed poorly in subsequent draws.
Table II summarizes part of the first process.
As the table shows, the measured and calculated draw
stresses for both first and second draws arc in good
agreement. However, internal stresses, calculated from
Eqs. (27) and (18), arc low on the first pass but high
on the second pass. This is a direct result of the
high A values produced by using the existing dies with
a = 6 ° . Open circuits of the Chromel thermoelement
occurred on the second draw for a -to-o ratios of
108.4 and 99.8%. As development continued and the
a , -to-a ratio for the second draw was reduced by
dfc. e
lowering a (and thus A ) , the incidence of open circuits
decreased. All failures were caused by Chromel thermo-
element opens; no sheath breaks occurred.
Table III summarizes the second and third draw-
anneal processes. The second process used draw dies
with a semiangle of 11°, based on the predictions of
Fig. 9, and an attempt was made to limit die-bearing
length to 15% of L (Fig. 6 ) . Correlations between
measured and calculated draw stresses decreased consi-
derably with an additional 10 to 15% first-draw cold
work. First-draw predicted values average 17% below
those measured. Extrapolation of the strain-hardening
constants to high cold work values or frictional behav-
ior that results in the friction factor f increasing
with cold work [Eq. (9)] could account for this.
Process III, presented in Table III, summarizes
the latest draw-anneal schedule, still under devel-
opment, in which 55% cold work between anneals is
achieved by drawing in two draws. Dies for the second-
draw pass were designed with a - 3° in accordance with
the predictions of Fig. 10. The measured first-draw
stresses were also higher than those predicted by Eq.
(3), but the measured second-draw stresses were lower
in contrast to those of process I with lower first-draw
cold work.
Errors in the strain-hardening constant n, the
redundant work term $, or the friction factor f at high
first-draw cold work must have occurred to account for
these discrepancies. Second-draw redundant work is
high for process I and low for process III, so higher
redundant errors would be expected in process I than in
III if $ errors dominate. Some strain-hardening error
at high cold work is probable. However, draw tests at
e = 0.42 that resulted in sheath breakage (a = a, )
e di
indicated that a is within about + 8 % of the value
obtained from thermocouple strain-hardening constants.
Since the apparent coefficient of friction u is higher
for thermocouple drawing than for wire drawing, it is
reasonable to assume that an additional increase in u
at high cold work accounts for much of the discrepancy
in calculated vs measured draw stress. For this as-
sumption to be true, u, which is constant at 0.19 for
cold work up to 30%, must increase sharply to u = 0.3
for cold work above 40%. No mechanism is yet available
to account for this phenomena.
The schedule developed in phase I was used to pro-
duce thermocouple bulk material with an average length
of about 50 m. Failure of the Chromel thermoelement
caused by high internal stresses was the limiting fac-
tor. The phase II schedule was used to produce thermo-
couple bulk material with an average length of about
100 m with sheath breakage limiting the process. Al-
though long bulk material lengths are possible with
process III above about 1-mm diameter, development has
not been completed to obtain 0.5-mm material in lengths
comparable with those of process II.
First-draw cold work for both processes II and III
is limited to a maximum of about 42% by sheath break-
age. The process III second draw, while extending the
cold work up to 20%, is limited on the low side of o
by friction and on the high side by high internal
stresses. When two draws are used to obtain maximum
cold work, diametrical tolerances of both draw dies are
critical to maintaining second-draw cold work within 10
to 20?o as thermocouple diameter is reduced below about
1 .5 mm.
More than 3500 m of bulk thermocouple material was
produced during process development. Figure 12 shows
the microstructure of the Chromel and Alumel thermoele-
ments resulting from an average of 40% cold work be-
tween anneals. The micrograin sizes of the thermoele-
ments are ASTM 11 and 9 (15,900 and 3,970 grains/mm2),
respectively.

SUMMARY AND CONCLUSIONS

Premature failure of small-diameter, type K Inco-


^el-sheathed thermocouples required an investigation
into the causes of failure and ways to improve life-
time. Failure was judged to be caused by thermoelement
grain growth and embrittlemenl. The brittle condition
was characterized by large grains with respect to ther-
moelement diameter and by voids and oxides formed along
the grain boundaries. Major contributors to this con-
dition were impurities introduced during thermocouple
assembly and, particularly, critical amounts of cold
work left in the thermocouple because of incomplete
annealing or introduced by swaging during FRS fabrica-
tion. The embrittled thermocouples subsequently failed
when thermally activated expansion mismatched between
them, and the FRS cladding imposed relatively small
axial strains on the thermoelements, causing grain-
boundary separations. Three characteristics were es-
tablished as important to improved thermocouple per-
formance under high-temperature; thermal cycling con-
ditions :
1. elimination of prestrain in the thermoelements
prior to operation of the thermocouple above their
recrystal1ization temperature;
2. minimization of oxygen, moisture, and hydroxides
within the thermocouple; and
3. refinement of the thermoelement grain size to meet
AST}) micrograin size 9 or smaller (as a goal).

Significant improvements in thermocoup'e perfor-


mance were demonstrated simply by eliminating small
amounts of cold work in the material by a recovery an-
neal prior to their use. Additionally, a 1000°C < 10 ^i
bakeout of the insulators prior to their assembly into
the thermoelement sheath, careful cleaning, and thermo-
couple assembly with an inert cover gas ve;e added to
the thermocouple fabrication process. Finally, refine-
ment of thermoelement grain size to ASTM micrograin
size 9 for Alumel and 11 for Chromel was accom-
plished by development of a new draw-anneal process.
Annealing conditions were established as 1025°C
for 30 to 120 s depending on thermocouple diameter.
The drawing process was optimized by application of the
general wire-drawing equation a = f<J>/ade to thermocou-
ple drawing. The equation was extended to describe
internal stresses for multiple-pass drawing of thermo-
couples, and process parameters were evaluated rr.d op-
timized to increase cold work between anneals from 25
to 6O7o. This maximum cold work level was achieved in
two draws between anneals averaging 40 and 15%, respec-
tively. High second-draw yield stress caused by first-
draw strain hardening combined with frictional losses
to limit second-draw cold work to 10 to 20"!b with a draw
die semiangle of 3 to 4 ° . The friction factor f = (1 +
H cot a) was found to be about 25% higher for thermo-
couple drawing than for equivalent wire drawing. Jt
also appeared to be a function of the per-draw cold
work level when cold work was >30%.
Improved grain size was most easily achieved by
single-pass drawing to an average of 4 0 % cold work be-
tween anneals. Although further improvement in grain
size occurred with the 55%, multiple-pass process, dia-
metrical tolerances were critical below about 1.5 mm
because small diametrical changes in either draw die
caused a relatively large second-draw cold work
change.
Table 1. FRS thermocouple requirements

Type K (Chromel-Alumel)
Sheath Inconel 600
Insulation MgO
Junction Insulated
Length, mm 1550 to 2750
Diameter, mm 0.51
Lifetime, h 1000 minimum
Time at 1100°C, h 40
Time above 1300°C, min -10
Thermal cycles 2000 minimum
Cycling range, (°C)
Normal 300 to 800
<10 cycles 300 to 1400
Ramp rate, °C/s 1 to 50
Accuracy, °C +8 to 800
+15 from 800 to 1400
Table II. Type K, Inconel-sheathed thermocouple draw-anneal schedule - prscess I

Percent cold Calculated Process


Calculated Measured
work ife) Anneal s Error internal a,r/cr x 100 parameters
draw stress draw stress dE E
at 1025°C stress
ad (kg/m*) <id (kg/m 1 ) (% )
per draw total <TJT, (kg/m2) a A

26.4 58.6 59.3 -1.2 66.0 69 .9 6 1.6


15.3 41.7 90 52.7 51.9 +C.6 100.1 93 .3 6 2.7
30.5 68.9 69.5 +0.6 74.1 75 .4 6 1.4
12.2 42.7 60 45.5 45.7 -0.2 104.3 96 .6 6 3.4
28.1 62.8 60.9 +3.1 66.5 69 .3 6 1.5
9.7 37.8 45 40.4 50.8 -20.5 113.1 108 .4 5 3.6
27.6 61.4 61.6 -0.3 67.2 70 .3 6 1.5
12.5 40.1 45 45.6 47.8 -2.5 105.9 99 .8 6 3.4
28.6 64.2 64.6 0.6 70.2 72 .7 6 1.5
11.5 40.1 30 50.6 46.7 +8.4 89.0 83 .9 4 2.4
Table III. Type K, Inconel-sheathed thermocouple draw-anneal schedule — processes II and III

Percent cold Calculated Process


Calculated Measured
work (e) Anneal s Error internal a,J<s x 100 parameters
draw stress dE e
per draw total
at 1UZ5°(, o (kg/m2)
°d(k8/nJ) draw stress (%) a A

-
Fhase II
40.8 90 74.4 84.7 -12.2 94.2 88.4 11 1.9
40.8 60 74.4 82.8 -10.1 92.3 77.7 11 1.9
37.8 45 68.2 86.4 -21.1 96.6 92.6 11 2.0
39.3 45 71.7 87.3 -17.9 97.5 92.4 11 2.0
41.8 30 76.0 99.6 -23.7 108.0 100.5 11 .1.8
Phase III
41.5 76.1 89.9 -15.4 99.4 92.8 11 1.9
15.9 57.4 60 85.5 70.5 +21.3 90.7 77.3 3 1.3
42.5 77.6 101.7 -23.7 110.1 102.1 11 1.8
18.6 61.1 45 98.9 90.0 +9.9 101.6 85.1 3 1.1
43.4 79.7 110.6 -27.9 119.0 109.7 11 1.8
17.8 61.2 30 96.0 92.1 +4.2 108.1 90.5 3 1.2
MI'«EH. •COOLING AIR

COOLING A I R .

l ^—-HtftT
HEAT EXCHANGER r& V. - /< " EXCHANGED
NUMBER 2 v
NUMBER 1

VALVE FCV-3«
• •J
/• o
TEST ^ '
VESSEL I
•; •*»•

N,1-'.;!-^
CIRCULATOR
- G R O Y I O C FLANGE
MOTOH REMOVAI (TYPICAll
FIXTURE OEPRESSURIZATION
SYSTEM

Fig. 1. Core flow test loop.


ao° T yp
CENTRAL HN iNSULATlQN */*****(*. \
NICKEL ROD

316S5 SHEATH
— 0S1 mm J
INTERNAL THERMOCOUPLES ENLARGED
SECTION A-A TMtBMOCOtJPU

- 1220 mm HOUGHtNEO ClAO •» fllCKFL TUBE

/ I1 COPPER TUBE

THEOMOCOUPLS

. :_ ~*3

Fig. 2. GCFR-CPTL FP.S.


- 20 - INCONFL 600

316 SS
1 5
O
CHROMEL
< 10
x
0 5
c
INCONEL 601 AND CHHOMEL
ZOO ARC APPROXIMATELY
FUUIVAI I NT
-0!
1000 2000
TEMPERATURE ("Cl

Tig. 3. Average linesr thermal expansion of ma-


terials for CFTL FRS cladding and thermocouple compo-
nents .
0.12

J 0.10
J
j
; cos
CRITICAL AMOUNT
| 0.06 _ I OF COLD-WORK FOR —
I
I
I RECRYSTALLIZATION
j 0.04
INITIAL
>0.02 GRAIN SIZE
0.05 mm
o
LU
11
0.00
30 40 50
COLDWORK (%)

Fig. 4. Schematic curves of recrystallized grain


size vs amount of cold work before recrystallization.
Fig. 5. Photomicrographs of Alumcl tlu-rmoclemcnt
from (a) a swaged FRS and (b) a nonswagcd FRS, both of
which endured -1000 h of operation with more than 2000
thermal cycles. Voids in (a) render it very brittle
while the void-free sample of (b) is ductile (original
reduced 8 % ) .
A,

I:ig. 6. The steady-state ornving operation.


STRAIN

F i g . 7 . F l o w - c u r v e ( K E ) for a solid w i r e that


e x h i b i t s s t r a i n h a r d e n i n g as well as work curve for
ideal (/ode) and typical (1/ti / c J e ) d r a w i n g p r o c e s s .
la)

Fig. 8. A schematic representation of radial


stress and strain pattern across a thermocouple drawn
with A = 3 and e = 0.17. Average strain is increased
to ~E by nonuniform distribution, and strain of thermo-
element is about 0.8c. Internal stress is maximum in
the center and depends on yield stress as well as A.
ORNL-DWG 81-16303 ETD
4.0

6 8 10 12 14
DRAW DIE SEMIANGLE - a (deg)

Fig. 9. $ff product vs draw—die scmiangle at va-


rious amounts oC cold work (n = 0.19).
ORNL-DWG 81-16304 ETD

8.0

3 4 5
DIE SEMIANGLE - a (deg)

Fig. 30. 'tfy product vs draw-die semiangle for


various amounts of second-draw cold work (fi = 0.19).
125
I I I 1
INCONEL
THERMOCOUPLE^.


100

^^.——''CHROMEL
o
X —
75

I-
50 —

INCONEL
LLJ THERMOCOUPLE ALUMEL CHROMEL
a: 25 — a - kc
n
k X 10"6 137.0 127.0 117.0 —
n 0.28 0.28 0.30

9R 1 1 I

0.1 0.2 0.3 0.4 0.5


TRUE STRAIN - (mm/mm)

Fig. 1 1 . True stress vs true strain for an ln-


concl-sheathed Type K thermocouple and for Chromcl and
Alurnel thermoelements.
-90M-

Fig. 12. Chromel (left) and Alumel average grain


size is 9 and 17 n (ASTM 11 and 9 ) , respectively, after
grain size refinement (400x). (Original reduced 16%).
REFERENCES

"Research sponsored by the Office of Research and


Technology, U.S. Department of Energy under con-
tract \V-74050-eng-26 with the Union Carbide Cor-
poration—Nuclear Division.

Permanent address: Oak Ridge National Laboratory,


P.O. Box X, Oak Ridge, Tennessee 37830.

Permanent address: Y-12 Development Division,


P.O. Box Y, Oak Ridge, Tennessee 37830.

1. Potts, J. F., Jr., and McElroy, D. L., Thermocou-


ple Research to 1000°C ~ Final Report November 1,
1957 through June SC, 1959, ORNL-2773, Oak Ridge
National Laboratory, 1969.

2. Anderson, R. L., and Kollie, T. G., Accuracy of


Small Diameter Sheathed Thermocouples for the Core
Flow Test Loop, ORNL-5401, Oak Ridge National
Laboratory, 1979.

3. Reed-IIill, R. E., Physical Metallurgy Principles,


Van Nostrand Reinhold Company, 1964, p. 144.

4. Gat, Uri, and Kasten, P. R. , Gas-Cooled Reactor


Program'8 Annual Progress Report for Period Ending
December SI, 1978, ORNL-5560, Oak Ridge National
Laboratory, p. 150.

5. Spooner, N. F., and Thomas, J. M., "longer Life


for Chromel-Alumel Thermocouples," Met. Prog.
(November 1955).

6. Shewman, P. G., Transformations in Metals, McGraw-


Hill Book Company, New York, 1969, Chap. 3, pp.
69-125.

7. Gat, Uri, and Kasten, P. R., Gas-Cooled Reactor


Programs Annual Progress Report for Period Ending
December 31, 1978, ORNL-5560, Oak Ridge National
Laboratory, pp. 140-62.

8. Ludwig, R. I,., private communication to Anderson,


R. L., Jan. 15, 1979; Subject: Examination of
Thermal Cycled Type K Sheathed Thermocouples.

9. Dyson, B. F., and llenn, D. E., "The Effect of Room


Temperature Prestrain on Grain Boundary Cavitation
in Nimonic 80A," J. Micros., 97 (1/2), 165-70
(January-March 1973).
10. Dyson, B. F. , and Rogers, M. J., "Prestrain, Cavi-
tation, and Creep Ductility," Met. Sci. J. 8, 261~
66 (1974).

11. Backofen, W. A., Deformation Processing, Addison-


Wesley Publishing Co.. Inc., Reading, Mass., 1972,
pp. 190-196.

12. Backofen, W. A., Deformation Processing, Addison-


Wesley Publishing Co., Inc., Reading, Mass., 1972,
pp. 227-228.

13. Backofen, W. A., Deformation Processing, Addison-


Wesley Publishing Co., Inc., Reading, Mass., 1972,
pp. 13 8-13 9.

14. Hill, R., The Mathematical Theory of Plasticity3


Oxford University Press, Anon House, London,
England, 1964, p. 257.

15. Backofen, W. A., Deformation Processing, Addison-


Wesley Publishing Co., Inc., Reading, Mass., 1972,
pp. 145-147.

16. Backofen, W. A., Deformation Processing, Addison-


Wesley Publishing Co., Inc., Reading, Mass., 1972,
p. 140.

You might also like