OI{NL

t;1ASTER COpy
OAK RIDGE NATIONAL LABORATORY

•
operated by
UNION CARBIDE CORPORATION
for the
U.S. ATOMIC ENERGY COMMISSION

ORNL TM- 274.jJC 7'

,.)0 t:

THE DETECTION OF BOILING

IN A WATER-COOLED NUCLEAR REACTOR

A. L. Colomb
F. 1. Binford

NOTICE

This document contains information of a preliminary nature and was prepared

primarily for internal use at the Oak Ridge Notional Laboratory. It is subject
to revi sion or correction and therefore does not represent a final report. The
information is not to be abstracted, reprinted or otherwise given publ ic dis·
semination without the approval of the ORNL potent branch, Legal and Infor·
mation Control Department.
r-----------------------------lEGAlNOTICE----------------------------~

report was prepared as an account of Government sponsored work. Neither the United States,
Commission , nor any porson acting on behalf of the Commfiision:
any worronty or representotion l expressed or implied, with respect to the accuracy,
completeness, or usefulness of the information contoined ;n this report, or thot the use of
any information, oPPOr'otu5 , me1hod, or process disclosed in this report may not infringe
privately owned fights; or
B. Assumes any liabilities with respect to the use of, or for damages resulting ftom the use of
ony information, apparatus, method, or process disclosed in .h!,s report.
As used in the above, "person acting on behalf of the Commiss.ion" includes ony employee or
contractor of the Commission, Of employee of such controctor, to the extent that such employee
or controctor of the Commission l or employee of such contractor prepares, disseminates, or
provides access to, any informotion pursuant to his employment or contract with the Commission~
or his employment with such controctor.
 ORNL-TM-274

Contract No. W-7405-eng-26

THE DETECTION OF BOILING

IN A WATER-COOLED NUCLEAR REACTOR

A. L Colomb
F. T. Binford

Paper presented at Conference on

Light-Water-Moderated Research Reactors
June 11-14, 1962
Gatlinburg, Tennessee

Date Issued
AUG 17 1962

OAK RIDGE NATIONAL LABORATORY

Oak Ridge " Tennessee
operated by
UNION CARBIDE CORPORATION
for the
U. S. Atomic Energy Commission
 THE DETECTION OF BOILING
IN A WATER-COOLED NUCLEAR REACTOR

A. L. Colomb and F. T. Binford

ABSTRACT

This paper will describe measurements made at ORNL to study the

feasibility of boiling detection in a water-cooled nuclear reactor.
The methods selected for the detection of boiling are:

1. Measurement of the acoustical noise produced by the generation

of bubbles.
2. Measurement of changes in the reactor-power spectral density
produced by bubbles.

Preliminary results indicating that both methods could detect

boiling are shown.
 ··2 -

INTRODUCTION

One of the mechanisms most 1 ike ly to cau~e damage to a watet>C('o }.,d

reactor of the ORR type is overheating of the fuel plates resulting in
boiling burnout. The burnout phenomenon has be~n described in the
l
literature ,2 and is illustrated graphically in Figure 1. In this figurc:>
the heat flux is plotted against oT, the difference between the tempeJ:3tu.e
of the heated surface and the saturation temperature of the bulk fluid.
The region AB repr.esents that part of the curve where heat transfer ,o<;curs
by forced convection. The s11rface temperature at B may be slightly above
the saturation temperature. At B boiling commences at the heated surface.
Initially, bubbles arise from a few points of nue leation. These nuc leatioll
points become more numerous as the heat flux is raIsed. In the region Be
this nuc Ie ate boiling enhances the heat trans fer because of the heat of
vaporization required to cause formation of the bubble. This e.ffect
results in an increase in the slope of the curve as shown. As the heat
flux is further increased above C, the bubbles begin to bec.ome so l1
'
1merou5
that they interfere with one another. This coalescing of the bubbles
results in the formation of a blanket or film of steam on the heat.,t':Hn3fet
surface which insulates it and forces the surface temperature highe~.

Finally, when the heat flux is increased to D, the stearn film i.s so
effective in insulating the fuel surface that the temperature rises shncply
even if there is no increase in heat flux and, in fact, will usually C'. Hl,

tinue to rise despite a rather s izab Ie decrease iu heat fV1X.

Strictly speaking, the term "burnout II applies to that poirlt '.'In ti:rt

curve at which melting or serious structural damage to the fuel IJCCUt'L~.

This may actually lie somewhere between E and F; however, since the
temperature continues to rise even with a decrease in heat flux onc.e th"
point D is reached, it seems reasonable to assume that damage will N::Cllr

in the reactor if the point U is reached. This puint will, accl)td1>.lg1y,

called the "burnout point".
Reactors such as the ORR, 1.ITR, and HFIR are des igned to operau~ In
the forced convection region AB. However, p~'uvided the magrlH,ude- .,·f Ll,<~

void coefficients are small and the void volume ls also small, no SeL'Cl!S

difficulty should be encountered even if the heat flux is great enough t.C:i
 UNCLASSIFIED
ORNL-LR- DWG 61401A
i07 r---------.--------.--------~--------.~

i06
-....
.c:.
•
N
-
....
..........

CD-
..-
::J

105
x
::::>
..J
LL

~
W
J:
" . 40 4 A

10 3 L -_ _ _ _ _ _ _ _~_ _ _ _ _ _~_ _ _ _ _ _ _ _~_ _ _ _ _ _ _ _~

Relationship Between Heat Flux and Temperature

Difference Showing the Various Regimes of Boilingg
Fig. I.
 -4-

cause some portions of the core to operate in the nucleate boiling region
BC. In fact, the most efficient use of the cooling system is realized in
this region.
The precise shape of the heat f1ux-~T curve will depend upon a number
of parameters including the pressure, the coolant velocity, the subcoo1ing,
and the hydraulic diameter. Correlations exist'3 4 which permit the pre-

diction of the burnout heat flux fairly accurately provided these pa-
rameters are known. While it is possible to know the nominal value of
these parameters, it is very difficult, from a practical standpoint, to
know them with sufficient accuracy at every point in the core. Moreover,
because of local flux gradients and irregularities in fuel density, it is
not possible to predict the exact heat flux at every point in the core.
This combination of uncertainties, usually grouped together under the
heading of "hot spot factor", opens up the possibility that even though the
reactor is designed to operate well down in the forced convection region
certain portions of it may be operating in or near the nucleate boiling
region. It follows that if the heat flux of the reactor is increased until
it is operating nominally in the nucleate boiling region, certain portions
of it may be operating dangerously close to burnout. For this reason both
the ORR and the LITR are operated well below the point at which nucleate
boiling is expected to occur.
In addition to the uncertainties mentioned above, there exists the
distinct possibility that the blocking of a fuel channel by a foreign
object in the ORR might sufficiently restrict the coolant flow to permit
burnout to occur. Moreover, it is recognized that the inadvertant reversal
of one of the nonuniform1y loaded fuel plates during manufacture would very
probably lead to burnout in the HFIR core if that core were brought to full
power. For these reasons, it is considered important to develop a device
which will warn of the approach to burnout conditions and thus enable
corrective measures to be taken before any serious damage is done.
The most promising procedure appears to be the development of a device
which will detect the onset of boiling. It would be highly desirable if
the device could differentiate between nucleate boiling and film boiling;
i.e., could indicate whether or not the condition Be or the condition
 -5-

CD is obtained. During the initial tests of the ORR, boiling was

detected by means of a sensitive fast recorder connected to a neutron
5
monitoring device. The indication of boiling was a characteristic
fluctuation of the neutron current reaching the monitorin6 device because
of the variation in moderator density due to bubble formation. In this
case, what was actually detected was probably the evolution of non-
condensible gases from the heated water; however, the indications of
boiling occurred under approximately the conditions predicted by theory.
An attempt to refine this method and to differentiate between nucleate
6
and film boiling at the WTR was terminated abruptly by the burnout of
a fuel element.
It seems clear that boiling can be detected by observing fluctu-
ations in the neutron density. It is not clear at this point how well
developed the boiling must be in order to be observed nor whether what is
seen is extensive nucleate boiling or a small region of bulk boiling.
This paper will describe the experimental program now being carried
out at ORNL to develop an instrument capable of detecting the onset of
boiling in a water-cooled nuclear reactor.
Two methods of detection are being investigated. The first one
already mentioned above consists of measuring the spectral density of t.1e
reactor power fluctuations to find out if the reactivity variations pro-
duced by the boiling bubbles can be observed and used as indication for
the amount of bubbles generated.
The second method is completely independent of the nuclear character-
istics of the reactor and could be used in any system where boiling occurs
on a heated surface. It is an acoustical method consisting of investi-
gation of the acoustical noise generated by the boiling process to deter-
mine if it can be discriminated in the presence of the general background
noise produced by the reactor hydraulic system.
In order to test these two methods quantitatively, some boiling has
to be generated in the reactor, the ORR. This can be done in two ways,
nuclear heating or electrical heating,
The second method was chosen because, although quite involved, it
has the advantage of producing at will, and without changing the reactor
power, a controllable amount of boiling at a well-defined position in the
reactor.
 -6-

BOILING GENERATOR
An instrumented electrical heater was built to generate boiling in
2
the reactor core. The heater consists of a lin. ,50-mil thick nichrome
plate connected to a 6-kw, AC power supply by large aluminum electrodes.
The heater is assembled in a core box having the same external dimensions
as an ORR fuel element and, therefore, can be accurately positioned in
the core.
The power generated by the heater section can be computed from
measurements of the total circuit current and of the voltage drop across
the nichrome plate.
The heater wall temperature is measured by three chromel-alumel
thermocouples brazed on the plate center line.
In order to achieve boiling at a reasonable power, the water flow
along the heater faces had to be limited by a calibrated orUBce to five
ft/sec at maximum reactor flow.
The boiling generator was first tested without forced cooling.
Figure 2 shows a wall temperature versus power measurement definitely
indicating the change in heat-transfer characteristics due to the onset
of boiling at approximately 1 kw. No attempts were made to compare this
measurement with existing correlations because, under natural convection
conditions, the coolant flow moves in a direction opposite to that for
which the orifice was designed; and the coolant flow cannot become well
established.
The measurements of wall temperature versus power for forced con-
vection are reported in Figure 3. There, the behavior of the measured
points compare well with the calculated curve. This curve was calculated
by using Colburn's correlation for a flow rate of five ft/sec and a
pressure of 30 psia. The almost constant discrepancy between the measured
and the calculated curves is ascribed to a systematic error in the measure-
ment of heater power. The voltage drop measured across the heater is equal
to the drop across the nichrome plate plus the drop across the nichrome-
aluminum contacts. The heat generated at this contact is certainly conducted
away by the aluminum; and, therefore, the power generated in the nichrome
will be smaller than that indicated by the instruments. This measurement was
 -7-
UNCLASSIFIED
ORNL- LR- DWG 68531
3

0
2
(

g0
C P
-- O.B
~
3::
1
nY
nr
0:
w y
3:
0 /0
a.. 0.6
0:
w
i/O
l-
e:(
w 0.4
J:
/ p
0

/n
/,
Va
0.2
0

0.1
50 75 100 125 150
A f, SURFACE TEMPERATURE MINUS BULK TEMPERATURE (OF)

Measured Temperature vs Power Curve for Natural

Convection.
Fig. 2.
 UNCLASSIFIED
ORNL-LR-DWG 68530
2

- ~ 1
-ffi
..:III::

0.8
~
0
a. 0.6 .

/l7~I
I
CD

~
I
'COLBURN

0.4
1 7 5 ft/sec, 30 psia
:::t

0.2 ~' ____~______~____~_____ L_ _ _ _~L_~

50 75 100 125 150 175
flt, SURFACE TEMPERATURE MINUS BULK TEMPERATURE (OF)

Measured Temperature vs Power Curve for Forced Con--

vection.
Fig. 3 ..
 -9-

interrupted before boiling was achieved by the failure of the thermocoup1e&

It was found later, upon examination of the heater in a hot cell, that the
nichrome-aluminum contact at one end of the heater must have had too high
a resistance and therefore produced a hot spot. The effect of this hot
spot was first to burn the thermocouple leads that passed very close to
it and, after approximately one week of operations, to burn the nichrome
at the contact, resulting in a complete failure of the heater.
The thermocouple failure made any qualitative measurements of boiling
effects impossible. Nevertheless, it was possible to estimate, with the
existing experimental values and the available correlations, that the
onset of boiling would occur between 2 and 3 kw.
The partial results presented in the next paragraphs were obtained
in the period of time between the failure of the thermocouples and the
heater failure.
ACOUSTICAL NOISE ANALYSIS

An underwater microphone was installed alongside the heater

electrodes at approximately 10 ft from the reactor core. The output of
this microphone was fed into a Hewlett-Packard wave analyzer and the
output of the wave analyzer was smoothed out by a passive integrator with
a 240-sec time constant.
This apparatus was used to measure acoustical spectra, and Figure 4
is a plot of the background noise and of the boiling spectrum in the
absence of reactor coolant flow showing that boiling produces a detect-
able amount of noise. Figure 5 indicates that the boiling noise varies
proportionally with the power generated in the heater. It is interesting
to note that the abrupt change in amplitude behavior occuring at 0.86 kw
corresponds fairly well with the value of 1 kw found previously (Figure 2)
for the onset of boiling.
Comparing Figure 4 with Figure 6, a plot of the hydraulic noise
measured with a reactor coolant flow of 18,000 gpm indicates that the
peaks measured in the boiling spectrum at 0.82 and 1.87 kilocycles could
be due to resonance effects in the reactor tank. The fact that two
different noise sources produced peaks at the same frequencies strongly
supports this interpretation. On the other hand, it appears that the
 -10-

UNCLASSIFIED
ORNL-LR-DWG 68528
10 I I I I I
NO FLOW IN REACTOR COOLING CIRCUIT
n
9
(

I, .I
--.... 8
en
o
>
- 7
II nI 1/\,

~
lLI
o
::>
i , \-.--WITH BOILING
f , " HEATER POWER 1.8 kw
t: 6
-I
I
a..
~
<I 5 II I\ \
lLI
en
o- IiI I I
I
iI \
z 4 r
-I
<I
U
i= 3
" ~
1111 , '(
\ i\
en
::>
o
~ 2
"~ H\
I
Ji\ !~ J \~
'\ !\
~ \/
I~'

1
)

\\ { \ -~
I'
...--t- BACKGROUND
\\{ \\
o 1 .....
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
FREQUENCY (kilocycles)

Acoustical Noise Spectrum; No Reactor Coolant Flow.

Fig. 4.
 UNCLASSIFIED
ORNL-LR-DWG 68525
9 .,
8
0;
-
~ 7
(5
V
/
- >

LUG
o
FREQUENCY 900 cps
_L
::::::>
t-
....J
Q. 5 /0
V
:E
«
.....
~4 .....
I
o
z
....J 3
«
f
u
t-
en
::::::>
ou
2 /
«
1
;,
..... - I/
o ~ .... -

o 0.5 1.0 1.5 2.0

POWER (kilowatts)

Acoustical Noise Amplitude as a Function of Heater Power.

Fig. 5.
 -12-
UNCLASSIFIED
ORNL-LR-DWG 68524
25
l3J
REACTOR COOLANT FLOW
18,000 gpm
20
)

-....
f I)

'0
->
lLJ
i
0
:.J
.... 15 I
...J
a.
:!
<t
J
I
lLJ
(J) I

c~
0
Z
(
...J
<t
10 (

U
....
(J)
:.J
0
U
<t

5( 1. ~
( ~
~
o
" ~)

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

FREQUENCY (kilocycles)

Acoustical Noise Spectrum; Reactor Coolant Flow t8,000 gpm.

Fig. 6.
 -13-

boiling spectrum peaks at 0.92, 2.07, and 3.17 kilocycles are character-
istics of boiling because they are located at frequencies where the
background noise has low values.
This series of observations indicates that at a pressure of 30 psia
the acoustical noise produced by boiling can be detected and that if the
background noise is not too great it can be detected at the onset of
boiling.
Nevertheless, this noise has a relatively small amplitude compared
with the amplitude of the hydraulical noise (Figure 6). An attempt was
made to discriminate the boiling noise against the hydraulical back-
ground noise in the region of the O.92-kilocycle boiling peak. Figure 7
is a plot of the noise amplitude with and without power in the boiling
generator. As one approaches the maximum of the peak the ratio of
boiling noise to background noise increases rapidly. The fact that the
boiling noise amplitude becomes smaller than the background noise at low
frequencies is ascribed to the effect of the magnetic field of the heater
electrodes on the microphone. This explanation still needs confirmation.
Unfortunately this test was interrupted by the failure of the boiling
generator before the maximum of the 0.92 kilocycle could be reached.
Future work with an improved boiling generator will use the in-
formation obtained up to the present to improve the sensitivity of the
detector. The microphone will be installed far away from the heater
conductors to avoid magnetic effects. The microphone will also be moved
as close to the reactor as radiation permits in order to increase the
boiling-noise-to-background-noise ratio. This ratio should be quite
sensitive to distance. The boiling noiseJbeing generated by a point sourc~

will decrease at least with the square of the distance between source and
detector. On the other hand, the background noise, being generated
throughout the system,will have a relatively constant effect on the
microphone wherever it is located.

POWER FLUCTUATIONS SPECTRAL DENSITY

To measure and analyze the reactor fluctuations a neutron sensitive

ionization chamber was placed in one of the standard chamber locations
close to the reactor core.
 UNCLASSIFIED
ORNL-LR-DWG 68526
5

w
o
::;:)
t-
...J
a.. _
~ 4
<t +-
(/)
w .-
c
en :::.J
o :>.
Z
...J~
<t:O
o"- NO HEATER POWER
-
.P-
I

(.) "-
i= E 3
en
::;:)
o
(.)
<t
HEATER POWER = 4.5 kw -

2
0.88 0.89
~.
0.90 0.91
FREQUENCY (kilocycles)

Boiling Noise Amplitude in the Vicinity of a Resonance.

Fig. 7.
 -15-

The current output of the chamber was 200 p amps and the total
amplitude of the current fluctuations 0.2 p amps. This indicates that
at a reactor power of 30 Mw the power fluctuations have an amplitude equal
to 0.1% of the operating power or 30 kw.
These current fluctuations are converted to a voltage on a 1 M-O
load resistor and then amplified by a factor of 100 in an AC coupled
amplifier having a flat frequency response from 0.1 to 1000 cps.
The amplifier signal is recorded on a tape recorder and digitalized
at a rate of 128 numbers per second by playing back the tape at a lower
speed than that used during the recording. The numbers obtained are
punched on an IBM card deck, and the auto-correlation function and the
spectral density of the signal are computed on a 7090 IBM computer.
The complete system was checked for frequency response by analyzing
the fluctuations produced by a gamma-ray source, and the spectral density
was found to be flat within the frequency range of interest.
Four 10-sec observations both with and without power on the heater
were taken and analyzed. The average auto-correlation functions obtained
from these samples are plotted in Figure 8. The standard deviation of
these curves is 22%. The cosine Fourier transform of the auto-correlation
functions (Le., the spectral densities of the samples) are plotted in
Figure 9. Here the standard deviations were found to be 45%.
Both the auto-correlation functions and the spectral densities
indicate a difference between the samples with boiling and those without
boiling, but the failure of the heater unit did not allow sufficient time
to obtain measurements sufficiently accurate to allow a good interpreta-
tion of the results. One conclusion can be drawn from this test. It
was found that for a given measurement duration the standard deviation
of the auto-correlation function is much smaller than that of the
spectral density. This indicates that a change in the reactor behavior
could be detected more rapidly from an observation of the auto-correlation
function than from an observation of the spectral density.
At this point, the results obtained,although not complete, indicate
that both methods, the acoustical noise analysis and the power fluctu-
ations analysis, could be used as a boiling-indicating device. It is not
 -16-

1.00
, UNCLASSIFIED
ORNl-lR-DWG 68529

z
0
-
I-
U
z 0.75
:J
u.
Z
0
-
I-
«
..J
~

~
w
a::
0:::
0 0.50
~
u
I
!--WITH BOILING
~
:J
« \\ /'HEATER POWER 4.5 kw

0
w
N
-'
« 0.25
\ V\ r
:?!
0:
0
Z
\ ~~ ''"--
/~-~
".~
"-.,r--.-L-
o
o 0.050 0.100 0.150 0.200 0.250
T, DELAY TIME (se,c)

Power Fluctuations Auto - Correlation Functions.

Fig. 8.
 UNCLASSIFIED
ORNL-LR-DWG 68527
1.0
-en
+-
c:
r-

::J

~ 0.8 I"

0
"-
WITH BOILING
+-
.0 HEATER POWER 4.5 kw" r
"-
S 0.6 1\ ~J
~

",~
I
>- I I
l- I
(/)
r I~ ,,-, I f-PI ...........a
I
Z
W 0.4 I"_J I
l\ r U- 1 ~J I
I I
L. eo

~.
I
0 I"_J ~l_ I- I
-l. r} u:: il
!"'"
I
- - --~ 1-
~
...J r" i- !-J
« ...
I
1--

a:: 0.2
I-
_tJ
u 1---1-- i-- l~
... [
W I
a.
en
P------- --- ~
o I
0.1 0.2 0.5 1 2 5 10 20 40
FREQUENCY (cps)

Power Fluctuations Spectral Densities.

Fig. 9.
 -18-

known what sensitivity could be achieved and what amount of time would be
required to obtain the desired information with sufficient accuracy.
An improved boiling generator is being built to pursue this in-
vestigation of the effects of boiling; and, as soon as sufficient knowledge
of this phenomenon is obtained, an instrument will be developed which is
capable of indicating when boiling occurs in a small part of the reactor
core.
 -19-

REFERENCES

1. W. H. MCAdams, Heat Transmission, 3d ed., p 370, McGraw-Hill,

New York, 1954.
2. M. Jakob, Heat Transfer, chapt 29 vol I, p 580 vol II, Wiley,
New York, 1949, 1957.
3. J. F. Wett, Jr., Nuclear Safety, ~(3), 10 (March 1961).
4. W. R. Gambill, Generalized Prediction of Burnout Heat Flux for
Flowing, Subcooled, Wetting Liquids - to be published.
5. J. A. Cox ~~, 0Berations Division Quarterly ReBort, ABril -
June, 1958, ORNL CF-58-6-122, (Oct. 13, 1958).
6. ReBort on WTR Fuel Element Failure ABril 2, 1960,WTR-49.
7. A. P. Colburn ~~, Effect of Local Boiling and Air Entrainment
on TemBeratures of Liquid Cooled Cylinder, U.S.N.A.C.A. Rpt
UD-NI, 1945; TN 1498, (March, 1948).
 Distribution

1-200. J. A. Cox

201. Document Reference Library

202-203. Central Research Library

204-206. Laboratory Records

207. ORO
208-222. Division of Technical Information Extension